This research focuses on developing persistence models for Rio Salado College. It is an initial effort to
predict persistence from one term to the next. Several ensemble models are experimented and compared in
their respective key metrics such as: confusion matrix, AUC, F1-Score, and feature importance.
Exploratory data analysis is undertaken to narrow the set of variables utilized in the models. Two models
were considered for possible implementation: a logistic regression and a gradient boosting machine. The
former is easier to implement and explain to non-technical personnel, while the latter behaves like a black
box. Based on key performance metrics, the model of choice was the gradient boosting machine.
Development and testing were conducted with python using jupyter notebooks. The author hopes that this
experimental process will fill a vacuum in the analytical needs of community colleges
PREDICTING SUCCESS: AN APPLICATION OF DATA MINING TECHNIQUES TO STUDENT OUTCOMESIJDKP
This project examines the effectiveness of applying machine learning techniques to the realm of college
student success, specifically with the intent of discovering and identifying those student characteristics and
factors that show the strongest predictive capability with regards to successful graduation. The student
data examined consists of first time freshmen and transfer students who matriculated at California State
University San Marcos in the period of Fall 2000 through Fall 2010 and who either graduated successfully
or discontinued their education. Operating on over 30,000 student observations, random forests are used
to determine the relative importance of the student characteristics with genetic algorithms to perform
feature selection and pruning. To improve the machine learning algorithm cross validated hyperparameter tuning was also implemented. Overall predictive strength is relatively high as measured by the
Matthews Correlation Coefficient, and both intuitive and novel features which provide support for the
learning model are explored.
Predicting Success : An Application of Data Mining Techniques to Student Outc...IJDKP
This project examines the effectiveness of applying machine learning techniques to the realm of college
student success, specifically with the intent of discovering and identifying those student characteristics and
factors that show the strongest predictive capability with regards to successful graduation. The student
data examined consists of first time freshmen and transfer students who matriculated at California State
University San Marcos in the period of Fall 2000 through Fall 2010 and who either graduated successfully
or discontinued their education. Operating on over 30,000 student observations, random forests are used
to determine the relative importance of the student characteristics with genetic algorithms to perform
feature selection and pruning. To improve the machine learning algorithm cross validated hyperparameter
tuning was also implemented. Overall predictive strength is relatively high as measured by the
Matthews Correlation Coefficient, and both intuitive and novel features which provide support for the
learning model are explored.
PREDICTING SUCCESS: AN APPLICATION OF DATA MINING TECHNIQUES TO STUDENT OUTCOMESIJDKP
This project examines the effectiveness of applying machine learning techniques to the realm of college
student success, specifically with the intent of discovering and identifying those student characteristics and
factors that show the strongest predictive capability with regards to successful graduation. The student
data examined consists of first time freshmen and transfer students who matriculated at California State
University San Marcos in the period of Fall 2000 through Fall 2010 and who either graduated successfully
or discontinued their education. Operating on over 30,000 student observations, random forests are used
to determine the relative importance of the student characteristics with genetic algorithms to perform
feature selection and pruning. To improve the machine learning algorithm cross validated hyperparameter tuning was also implemented. Overall predictive strength is relatively high as measured by the
Matthews Correlation Coefficient, and both intuitive and novel features which provide support for the
learning model are explored.
This document discusses a study that assessed the significance of corporate character scale in master's programs among students from different universities. The study surveyed 100 random students from 4 universities to understand their sentiments toward master's programs. Descriptive and inferential analyses were conducted to test hypotheses about correlations between variables and dimensions. The results found that respondents had positive dimensions of agreeableness and chic toward master's programs. It was concluded that most dimensions had a positive relationship with corporate character in master's programs.
A REVIEW OF AGENT-BASED SIMULATION FOR UNIVERSITY STUDENTS ADMISSIONCSEIJJournal
Multiple factors influence college selection and admission behaviors. Most researchers focused on the
academic and socioeconomic factors; the academic factors are high school GPA, SAT, admission tests, etc.
On the other hand, the socioeconomic factors could be family income and first-generation students, which
means parents did not complete their bachelor's degrees. However, some universities admission policies do
not pay any attention to the race or to the minorities even though some of them might be from the lowincome students which could not afford any admission tests, and they might lose their chance to get
admitted into their preferred universities. Therefore, most universities want a fairness admission system
that include both the disadvantaged students along with other high-score achievement students. Thus,
several simulations have been developed by using the agent-based models in order to simulate a real world
system by considering other factors and domains that are varied in the complexities. This paper aimed to
review several Agent-Based Models that are deployed by different admission offices from several
international universities and colleges around the world, which is classified based on the main contribution
of the simulations including the level of educational attainment as well as the universities selection
behaviors.
A Review of Agent-Based Simulation for University Students AdmissionCSEIJJournal
Multiple factors influence college selection and admission behaviors. Most researchers focused on the
academic and socioeconomic factors; the academic factors are high school GPA, SAT, admission tests, etc.
On the other hand, the socioeconomic factors could be family income and first-generation students, which
means parents did not complete their bachelor's degrees. However, some universities admission policies do
not pay any attention to the race or to the minorities even though some of them might be from the low-
income students which could not afford any admission tests, and they might lose their chance to get
admitted into their preferred universities.
Comparative Analysis of Naive Bayes and K-NN Approaches to Predict Timely Gra...anggarariocopasga
Graduation is a pivotal moment in higher education, significantly impacting institutional accreditation and public perception. This study aims to ensure that all students graduate punctually, recognizing the critical role of higher education in achieving this goal. Central to this effort are comprehensive datasets that capture academic performance throughout both undergraduate and graduate studies. These datasets include details such as university and program specifics, undergraduate and master’s GPAs, TOEFL scores, and the duration of study. By leveraging classification techniques within data mining, particularly K-NN and Naive Bayes, a comparative analysis was conducted to precisely predict the on-time completion of graduate students. The process of predicting graduation involves several stages, including data preprocessing, transformation, and the segmentation of data into training and testing sets. Subsequently, the selected methods are applied, and analyses are undertaken to accurately forecast graduation outcomes. Experimental findings reveal an 80% accuracy rate for Naive Bayes and 73% for K-NN. Notably, Naive Bayes demonstrates superior efficacy in predicting on-time graduation. However, to further refine accuracy, it is necessary to expand datasets and diversify the variables used in the analysis, such as incorporating additional academic and non-academic factors that may influence graduation timelines. The insights derived from this research hold significant implications for academic institutions, offering valuable guidance for implementing proactive measures to support students in completing their studies within the expected timeframe. By utilizing the findings of this study, educational institutions can develop tailored strategies and interventions to address potential barriers to timely graduation, such as enhancing academic advising, providing targeted support services, and optimizing course scheduling. These efforts will ultimately foster student success, improve institutional outcomes, and contribute to the overall excellence of higher education institutions.
AGENT-BASED SIMULATION FOR UNIVERSITY STUDENTS ADMISSION: MEDICAL COLLEGES IN...CSEIJJournal
Medical colleges are considered one of the most competitive schools compared to other university
departments. Most countries adopted the particular application process to ensure maximum fairness
between students. For example, in UK students apply through the UCAS system, and most of USA
universities use either Coalition App or Common App, on the other hand, some universities use their own
websites. In fact, a Unified Admission Application process is adopted in Jordan for allocating the students
to the public universities. However, the universities and colleges in Jordan are evaluating the applicants by
using merely the centralized system without considering the socioeconomics factor, as the high school GPA
is the essential player their selection mechanism. In this paper, the authors will use an Agent Based model
(ABM) to simulate different scenarios by using Netlogo software (v. 6.3). The authors used different
parameters such as the family-income and the high school GPA in order to maximize the utilities of the
fairness and equalities of universities admission. The model is simulated into different scenarios. For
instance, students with low family income and high GPA given them the priority in studying medicine
comparing with same high school GPA and higher family-income, as a results, after several rotations of
the simulation the reputation of medical schools are identified based on students’ preferences and seats’
allocated as it shows that high ranking universities are mainly allocated with have high cut-off GPA score.
PREDICTING SUCCESS: AN APPLICATION OF DATA MINING TECHNIQUES TO STUDENT OUTCOMESIJDKP
This project examines the effectiveness of applying machine learning techniques to the realm of college
student success, specifically with the intent of discovering and identifying those student characteristics and
factors that show the strongest predictive capability with regards to successful graduation. The student
data examined consists of first time freshmen and transfer students who matriculated at California State
University San Marcos in the period of Fall 2000 through Fall 2010 and who either graduated successfully
or discontinued their education. Operating on over 30,000 student observations, random forests are used
to determine the relative importance of the student characteristics with genetic algorithms to perform
feature selection and pruning. To improve the machine learning algorithm cross validated hyperparameter tuning was also implemented. Overall predictive strength is relatively high as measured by the
Matthews Correlation Coefficient, and both intuitive and novel features which provide support for the
learning model are explored.
Predicting Success : An Application of Data Mining Techniques to Student Outc...IJDKP
This project examines the effectiveness of applying machine learning techniques to the realm of college
student success, specifically with the intent of discovering and identifying those student characteristics and
factors that show the strongest predictive capability with regards to successful graduation. The student
data examined consists of first time freshmen and transfer students who matriculated at California State
University San Marcos in the period of Fall 2000 through Fall 2010 and who either graduated successfully
or discontinued their education. Operating on over 30,000 student observations, random forests are used
to determine the relative importance of the student characteristics with genetic algorithms to perform
feature selection and pruning. To improve the machine learning algorithm cross validated hyperparameter
tuning was also implemented. Overall predictive strength is relatively high as measured by the
Matthews Correlation Coefficient, and both intuitive and novel features which provide support for the
learning model are explored.
PREDICTING SUCCESS: AN APPLICATION OF DATA MINING TECHNIQUES TO STUDENT OUTCOMESIJDKP
This project examines the effectiveness of applying machine learning techniques to the realm of college
student success, specifically with the intent of discovering and identifying those student characteristics and
factors that show the strongest predictive capability with regards to successful graduation. The student
data examined consists of first time freshmen and transfer students who matriculated at California State
University San Marcos in the period of Fall 2000 through Fall 2010 and who either graduated successfully
or discontinued their education. Operating on over 30,000 student observations, random forests are used
to determine the relative importance of the student characteristics with genetic algorithms to perform
feature selection and pruning. To improve the machine learning algorithm cross validated hyperparameter tuning was also implemented. Overall predictive strength is relatively high as measured by the
Matthews Correlation Coefficient, and both intuitive and novel features which provide support for the
learning model are explored.
This document discusses a study that assessed the significance of corporate character scale in master's programs among students from different universities. The study surveyed 100 random students from 4 universities to understand their sentiments toward master's programs. Descriptive and inferential analyses were conducted to test hypotheses about correlations between variables and dimensions. The results found that respondents had positive dimensions of agreeableness and chic toward master's programs. It was concluded that most dimensions had a positive relationship with corporate character in master's programs.
A REVIEW OF AGENT-BASED SIMULATION FOR UNIVERSITY STUDENTS ADMISSIONCSEIJJournal
Multiple factors influence college selection and admission behaviors. Most researchers focused on the
academic and socioeconomic factors; the academic factors are high school GPA, SAT, admission tests, etc.
On the other hand, the socioeconomic factors could be family income and first-generation students, which
means parents did not complete their bachelor's degrees. However, some universities admission policies do
not pay any attention to the race or to the minorities even though some of them might be from the lowincome students which could not afford any admission tests, and they might lose their chance to get
admitted into their preferred universities. Therefore, most universities want a fairness admission system
that include both the disadvantaged students along with other high-score achievement students. Thus,
several simulations have been developed by using the agent-based models in order to simulate a real world
system by considering other factors and domains that are varied in the complexities. This paper aimed to
review several Agent-Based Models that are deployed by different admission offices from several
international universities and colleges around the world, which is classified based on the main contribution
of the simulations including the level of educational attainment as well as the universities selection
behaviors.
A Review of Agent-Based Simulation for University Students AdmissionCSEIJJournal
Multiple factors influence college selection and admission behaviors. Most researchers focused on the
academic and socioeconomic factors; the academic factors are high school GPA, SAT, admission tests, etc.
On the other hand, the socioeconomic factors could be family income and first-generation students, which
means parents did not complete their bachelor's degrees. However, some universities admission policies do
not pay any attention to the race or to the minorities even though some of them might be from the low-
income students which could not afford any admission tests, and they might lose their chance to get
admitted into their preferred universities.
Comparative Analysis of Naive Bayes and K-NN Approaches to Predict Timely Gra...anggarariocopasga
Graduation is a pivotal moment in higher education, significantly impacting institutional accreditation and public perception. This study aims to ensure that all students graduate punctually, recognizing the critical role of higher education in achieving this goal. Central to this effort are comprehensive datasets that capture academic performance throughout both undergraduate and graduate studies. These datasets include details such as university and program specifics, undergraduate and master’s GPAs, TOEFL scores, and the duration of study. By leveraging classification techniques within data mining, particularly K-NN and Naive Bayes, a comparative analysis was conducted to precisely predict the on-time completion of graduate students. The process of predicting graduation involves several stages, including data preprocessing, transformation, and the segmentation of data into training and testing sets. Subsequently, the selected methods are applied, and analyses are undertaken to accurately forecast graduation outcomes. Experimental findings reveal an 80% accuracy rate for Naive Bayes and 73% for K-NN. Notably, Naive Bayes demonstrates superior efficacy in predicting on-time graduation. However, to further refine accuracy, it is necessary to expand datasets and diversify the variables used in the analysis, such as incorporating additional academic and non-academic factors that may influence graduation timelines. The insights derived from this research hold significant implications for academic institutions, offering valuable guidance for implementing proactive measures to support students in completing their studies within the expected timeframe. By utilizing the findings of this study, educational institutions can develop tailored strategies and interventions to address potential barriers to timely graduation, such as enhancing academic advising, providing targeted support services, and optimizing course scheduling. These efforts will ultimately foster student success, improve institutional outcomes, and contribute to the overall excellence of higher education institutions.
AGENT-BASED SIMULATION FOR UNIVERSITY STUDENTS ADMISSION: MEDICAL COLLEGES IN...CSEIJJournal
Medical colleges are considered one of the most competitive schools compared to other university
departments. Most countries adopted the particular application process to ensure maximum fairness
between students. For example, in UK students apply through the UCAS system, and most of USA
universities use either Coalition App or Common App, on the other hand, some universities use their own
websites. In fact, a Unified Admission Application process is adopted in Jordan for allocating the students
to the public universities. However, the universities and colleges in Jordan are evaluating the applicants by
using merely the centralized system without considering the socioeconomics factor, as the high school GPA
is the essential player their selection mechanism. In this paper, the authors will use an Agent Based model
(ABM) to simulate different scenarios by using Netlogo software (v. 6.3). The authors used different
parameters such as the family-income and the high school GPA in order to maximize the utilities of the
fairness and equalities of universities admission. The model is simulated into different scenarios. For
instance, students with low family income and high GPA given them the priority in studying medicine
comparing with same high school GPA and higher family-income, as a results, after several rotations of
the simulation the reputation of medical schools are identified based on students’ preferences and seats’
allocated as it shows that high ranking universities are mainly allocated with have high cut-off GPA score.
Agent-based Simulation for University Students Admission: Medical Colleges in...CSEIJJournal
This document describes an agent-based simulation model developed using NetLogo to examine fairness in university admissions processes. The model simulates medical college admissions in Jordan. It defines two agent types: students and medical colleges. Student agents have attributes like GPA, family income, and college preferences. College agents have attributes like seat capacity. The model runs matching simulations based on GPA and family income to prioritize students from lower-income families. It analyzes scenarios varying numbers of students, colleges, and seat capacities. The goal is to maximize fairness by considering socioeconomic factors in admissions beyond just GPA.
Dr. Kritsonis has traveled and lectured extensively throughout the United States and world-wide. Some international travels include Australia, New Zealand, Tasmania, Turkey, Italy, Greece, Monte Carlo, England, Holland, Denmark, Sweden, Finland, Russia, Estonia, Poland, Germany, Mexico, the Caribbean Islands, Mexico, Switzerland, Grand Cayman, Haiti, St. Maarten, St. John, St. Thomas, St. Croix, St. Lucia, Puerto Rico, Nassau, Freeport, Jamaica, Barbados, Martinique, Canada, Curacao, Costa Rico, Aruba, Venezuela, Panama, Bora Bora, Tahiti, Latvia, Spain, Honduras, and many more. He has been invited to lecture and serve as a guest professor at many universities across the nation and abroad.
Dr. Kritsonis has traveled and lectured extensively throughout the United States and world-wide. Some international travels include Australia, New Zealand, Tasmania, Turkey, Italy, Greece, Monte Carlo, England, Holland, Denmark, Sweden, Finland, Russia, Estonia, Poland, Germany, Mexico, the Caribbean Islands, Mexico, Switzerland, Grand Cayman, Haiti, St. Maarten, St. John, St. Thomas, St. Croix, St. Lucia, Puerto Rico, Nassau, Freeport, Jamaica, Barbados, Martinique, Canada, Curacao, Costa Rico, Aruba, Venezuela, Panama, Bora Bora, Tahiti, Latvia, Spain, Honduras, and many more. He has been invited to lecture and serve as a guest professor at many universities across the nation and abroad.
IRJET- A Conceptual Framework to Predict Academic Performance of Students usi...IRJET Journal
This document presents a conceptual framework for predicting student academic performance using classification algorithms. The framework uses factors like socioeconomic status, psychological attributes, cognitive attributes, and lifestyle to analyze student performance based on their semester GPA. The document proposes classifying student performance into three classes (first class, second class, third class) based on their first semester GPA. Various classification algorithms like Naive Bayes, random forest, and bagging are evaluated on the student data to identify the best model for predicting performance. The conceptual framework is intended to guide the development of a recommendation system that can help educational institutions identify at-risk students early and improve student outcomes.
Clustering Students of Computer in Terms of Level of ProgrammingEditor IJCATR
Educational data mining (EDM) is one of the applications of data mining. In educational data mining, there are two key domains, i.e. student domain and faculty domain. Different type of research work has been done in both domains.
In existing system the faculty performance has calculated on the basis of two parameters i.e. Student feedback and the result of student in that subject. In existing system we define two approaches one is multiple classifier approach and the other is a single classifier approach and comparing them, for relative evaluation of faculty performance using data mining
Techniques. In multiple classifier approach K-nearest neighbor (KNN) is used in first step and Rule based classification is used in the second step of classification while in single classifier approach only KNN is used in both steps of classification.
But in proposed system, I will analyse the faculty performance using 4 parameters i.e., student complaint about faculty, Student review feedback for faculty, students feedback, and students result etc.
For this proposed system I will be going to use opinion mining technique for analyzing performance of faculty and calculating score of each faculty.
THE USE OF COMPUTER-BASED LEARNING ASSESSMENT FOR PROFESSIONAL COURSES: A STR...IAEME Publication
Background/Objectives: While the increase in classroom technology, it is necessary to examine how assessment is administered through technology. The purpose of this study is to understand how students and faculty are perceived and examine the effectiveness of the computer-based assessment in professional education courses (Educational Technology) at Northern Iloilo Polytechnic State College, Iloilo, Philippines. Methods: The research design utilized in this study is mixed-method research. A computer-based assessment was utilized to assess students' performance in educational technology. This instrument was validated, and pilot tested to establish reliability. Each campus of NIPSC selected ten students of 70 as respondents during Academic Year 2016-2017. Frequency count, mean, standard deviation, and Wilcoxon signed-rank test were statistical tools used for data analyses. Findings: The study's finding showed a high score of students in the posttest ensured better performance of the students in educational technology. The increase in the posttest per performance level of the students was due to an accurate measure of what they have learned in educational technology. The majority of students users agreed that online assessment was fasters than the paper and pencil form. Also, users agreed that online assessment is contemporary and more systematic. They also stated that online assessment is consistent with the teaching style, but they are less anxious. Furthermore, according to faculty and students, ninety percent (90%) believed that computer-based assessment accurately measures what they are teaching and what they learned in school, respectively. Novelty: With the current situation that the education system is in new normal, computer-based learning is important in flexible learning. And assessment using technology is a great help to both faculty and students. Thus, state universities and colleges (SUCs) should adopt this innovation to help teaching and learning.
“E-SurveyCo MIT: A Electronic Survey of Community Perception, Awareness, Acce...CatleenMadayag
An educational institution's guiding principles are its vision, mission, goals, and objectives. The different
accreditation body such as Accrediting Agency of Chartered Colleges and Universities in the Philippines,
Inc. (AACCUP), believe that the degree to which VMGOs are attained is used to evaluate a university. Thus
this study was conducted to measure the stakeholder's perceptions, awareness, acceptance, and
understanding towards the Isabela State University VMGO. Lack of in-person access to participants and
timeliness have pushed researcher to use electronic surveys to connect to the stakeholders of the university
due to COVID-19 pandemic. The stakeholders of the University are the respondents of the study and it was
categorized into two namely the Internal and External stakeholders. Internal stakeholders are composed of
Administrators or faculty members, non-teaching staff, and students while External stakeholders are
Parents/guardians, alumni, and industries, linkages, or cooperating agencies. A descriptive research
methodology was used by the researcher. Data from the respondents were gathered through convenience
sampling. Result showed that majority of the respondents were very highly aware of the University vision,
mission, college goals and objectives and highly aware on the dissemination activities by the university.
They also had very high acceptance and understanding of the university vision, mission, and campus goals.
The stakeholders also satisfy to the clarity and consistency of the VMGO. In terms of the program
objectives, the respondents were very highly aware of their specific programs and they have a very high
acceptance on their specific program objectives. Study also shows that stakeholder’s perception the VMGO
of the university is congruent with activities, practices, projects, and operations. Based on the foregoing, it
is recommended that Isabela State University should continuously work for the awareness and acceptance
of the ISU VMGO. It is recommended that both the College and CGS make more efforts in disseminating
the goals and objective of the MIT program especially to the parents. Activities must be more focused to
make them feel that the activities of the school is geared towards the realization of the VMGO. The
researcher also proofs that using online survey methods can be particularly useful and successful during
the current pandemic thus the researcher encourages others to utilized also e-survey to measure the
Perception, Awareness, Acceptance and Understanding of the vision, mission, goals of the college and
objective of the program.
Accurate prediction and early identification of student at-risk of attrition are of high concern for higher
educational institutions (HEIs). It is of a great importance not only to the students but also to the
educational administrators and the institutions in the areas of improving academic quality and
efficient utilisation of the available resources for effective intervention. However, despite the different
frameworks and various models that researchers have used across institutions for predicting performance,
only negligible success has been recorded in terms of accuracy, efficiency and reduction of student
attrition. This has been attributed to the inadequate and selective use of variables for the predictive models.
AN INTEGRATED SYSTEM FRAMEWORK FOR PREDICTING STUDENTS’ ACADEMIC PERFORMANCE ...ijcsit
Accurate prediction and early identification of student at-risk of attrition are of high concern for higher educational institutions (HEIs). It is of a great importance not only to the students but also to the educational administrators and the institutions in the areas of improving academic quality and efficient utilisation of the available resources for effective intervention. However, despite the different frameworks and various models that researchers have used across institutions for predicting performance, only negligible success has been recorded in terms of accuracy, efficiency and reduction of student
attrition. This has been attributed to the inadequate and selective use of variables for the predictive models. This paper presents a multi-dimensional and an integrated system framework that involves considerable learners’ input and engagement in predicting their academic performance and intervention in HEIs. The purpose and functionality of the framework are to produce a comprehensive, unbiased and efficient way of predicting student performance that its implementation is based upon multi-sources data and database
system. It makes use of student demographic and learning management system (LMS) data from the institutional databases as well as the student psychosocial-personality (SPP) data from the survey collected from the student to predict performance. The proposed approach will be robust, generalizable, and possibly give a prediction at a higher level of accuracy that educational administrators can rely on for providing timely intervention to students.
Accurate prediction and early identification of student at-risk of attrition are of high concern for higher educational institutions (HEIs). It is of a great importance not only to the students but also to the educational administrators and the institutions in the areas of improving academic quality and efficient utilisation of the available resources for effective intervention. However, despite the different frameworks and various models that researchers have used across institutions for predicting performance, only negligible success has been recorded in terms of accuracy, efficiency and reduction of student attrition. This has been attributed to the inadequate and selective use of variables for the predictive models. This paper presents a multi-dimensional and an integrated system framework that involves considerable learners’ input and engagement in predicting their academic performance and intervention in HEIs. The purpose and functionality of the framework are to produce a comprehensive, unbiased and efficient way of predicting student performance that its implementation is based upon multi-sources data and database system. It makes use of student demographic and learning management system (LMS) data from the institutional databases as well as the student psychosocial-personality (SPP) data from the survey collected from the student to predict performance. The proposed approach will be robust, generalizable, and possibly give a prediction at a higher level of accuracy that educational administrators can rely on for providing timely intervention to students. --
Fuzzy Measurement of University Students Importance Indexes by Using Analytic...IRJESJOURNAL
Purpose: The purpose of this paper is to apply a Fuzzy Analytical Hierarchy Process model FAHP for estimating students' importance indexes problem, where the measures of students' attitudes and responses are often uncertain or difficult to determine by using non-fuzzy model. Fuzzy set theory treats a kind of uncertainty called fuzziness. It shows that the boundary of “yes” or “no” is ambiguous and appears in the meaning of words or included in the subjunctives or recognition of human beings. Design/methodology/approach: This paper adapts FAHP to analysis students' satisfaction for the services submitted by the university to enhance learning process and circumstance environment for students, this area of researches is still out of university importance. For the purposes of the survey, questionnaires were designed for all the factors which are affecting in students' satisfaction and most probably all these factors are taken according to their suitable priority. There were five main criterions in the questionnaires. Criterion one focuses on administrative university services Si, the second criterion concentrate with teaching and learning process Ti, while the third criterion is university climate and student activities Ai, the fourth criterion covers the administrative facilities Fi, and the fifth criterion is learning facilities students Li, were asked to measure their perceived experiences with those criterions. Each criterion contains some of sub criteria. Findings: The results illustrate that the quality of teachers T1, Opportunities for recreational activities A1, fair evaluation for students T6, and remedial support A3 are the most important indexes for students. The proposed model would help decision-makers to enhancement the quality of the services and may be adding other facility to make the university more attractive. Research limitations/implications: This paper makes some assumptions such as the number of sample are taken from seven colleges only and not cover all colleges in the university and concentrated about the final levels. Originality/value: This paper introduces fuzzy theory with AHP approach to the research of university services as customers in public firms and it has reached some valuable conclusions, which has opened up a new field of study in the universities services area
The study sought to establish the impact of electronic information services (EIS) on students and alumni of Glasgow Caledonian University. Questionnaires were distributed to current students and alumni to collect both quantitative and qualitative data on EIS usage and information literacy. The data was analyzed to compare results between groups and identify commonly used resources. It was found that respondents understood the concept of information literacy, which alumni felt gave them an advantage in their careers. Unemployed alumni were less engaged with information literacy skills. The study concluded that information literacy skills learned in university could be directly applied and useful in the workplace.
Predicting students’ intention to continue business courses on online platfor...Samsul Alam
The objective of this study was to analyze the intention of a University's business department students to continue their studies on e-learning platforms during the ongoing COVID-19 pandemic. To this end, a questionnaire was developed to collect primary data from students in business fields. The study took into account more than 285 respondents from two different universities and relied on the expectation confirmation model (ECM) theory and the structural equation model. The partial least squares (SEM-PLS) method was used to analyze the data. The results of the study showed that task skills (TS) and task challenges (TC) were significant for the enjoyment (EN) of the students which in turn had a positive effect on the satisfaction levels. Confirmation (CON) had an impact on the post adoption perceived usefulness (PAPU), which was deemed positive for student satisfaction (SAT). The SAT and psychological safety (PS) of online learning platforms were found to positively influence the continuance intention (CI) on e-learning platforms. Finally, both SAT and PS of online learning platforms were observed to positively influence CI on e-learning platforms. Further research in this area could be useful in making decisions about promoting educational programs based on e-learning. The researchers recommend that academicians and policymakers must ensure appropriate arrangements for teaching on e-learning platforms.
PREDICTING ACADEMIC MAJOR OF STUDENTS USING BAYESIAN NETWORKS TO THE CASE OF ...ijcax
In this study, which took place current year in the city of Maragheh in IRAN. Number of high school students in the fields of study: mathematics, Experimental Sciences, humanities, vocational, business and science were studied and compared. The purpose of this research is to predict the academic major of high school students using Bayesian networks. The effective factors have been used in academic major selection
for the first time as an effective indicator of Bayesian networks. Evaluation of Impacts of indicators on each other, discretization data and processing them was performed by GeNIe. The proper course would be advised for students to continue their education.
PREDICTING ACADEMIC MAJOR OF STUDENTS USING BAYESIAN NETWORKS TO THE CASE OF ...ijcax
In this study, which took place current year in the city of Maragheh in IRAN. Number of high school
students in the fields of study: mathematics, Experimental Sciences, humanities, vocational, business and
science were studied and compared. The purpose of this research is to predict the academic major of high
school students using Bayesian networks. The effective factors have been used in academic major selection
for the first time as an effective indicator of Bayesian networks. Evaluation of Impacts of indicators on
each other, discretization data and processing them was performed by GeNIe. The proper course would be
advised for students to continue their education.
PREDICTING ACADEMIC MAJOR OF STUDENTS USING BAYESIAN NETWORKS TO THE CASE OF ...ijcax
In this study, which took place current year in the city of Maragheh in IRAN. Number of high school students in the fields of study: mathematics, Experimental Sciences, humanities, vocational, business and
science were studied and compared. The purpose of this research is to predict the academic major of high school students using Bayesian networks. The effective factors have been used in academic major selection for the first time as an effective indicator of Bayesian networks. Evaluation of Impacts of indicators on each other, discretization data and processing them was performed by GeNIe. The proper course would be
advised for students to continue their education.
Predicting student performance in higher education using multi-regression modelsTELKOMNIKA JOURNAL
Supporting the goal of higher education to produce graduation who will be a professional leader is a crucial. Most of universities implement intelligent information system (IIS) to support in achieving their vision and mission. One of the features of IIS is student performance prediction. By implementing data mining model in IIS, this feature could precisely predict the student’ grade for their enrolled subjects. Moreover, it can recognize at-risk students and allow top educational management to take educative interventions in order to succeed academically. In this research, multi-regression model was proposed to build model for every student. In our model, learning management system (LMS) activity logs were computed. Based on the testing result on big students datasets, courses, and activities indicates that these models could improve the accuracy of prediction model by over 15%.
Performance of student is dependent on their subject select and faculty’s expertise who teach the subject. Sometimes subject selection done by the student. Few students’ select subject wisely, few of them select because of friend influence, faculty influence, without thinking on it. To reduce the cost and related overhead department tries to reduce the variety of elective. Wisely selected and offered subject will increase the performance of student and throughput of the faculty. In this paper a recommender system is proposed which find the score of knowledge level for student and score of faculty score. These scores will be used to recommend the right subject to adept faculty of that subject.
A Methodology To Analyze Self-Reflection In E-PortfoliosRick Vogel
This document presents a methodology for analyzing self-reflection in e-portfolios created by students in the Grand Challenge Scholars Program at the University of Maryland, Baltimore County. The methodology applies two existing frameworks - Robert Grossman's levels of reflection and AAC&U's Integrative Learning rubric - to assess self-reflection. An initial analysis of e-portfolios from the first two cohorts found students did not reach high levels of reflection with a generic prompt. The analysis is ongoing and expects to see deeper reflection in newer portfolios with changes to expectations and prompts. The goal is to evaluate the program and identify interventions to enhance students' learning experiences.
Agent-based Simulation for University Students Admission: Medical Colleges in...CSEIJJournal
This document describes an agent-based simulation model developed using NetLogo to examine fairness in university admissions processes. The model simulates medical college admissions in Jordan. It defines two agent types: students and medical colleges. Student agents have attributes like GPA, family income, and college preferences. College agents have attributes like seat capacity. The model runs matching simulations based on GPA and family income to prioritize students from lower-income families. It analyzes scenarios varying numbers of students, colleges, and seat capacities. The goal is to maximize fairness by considering socioeconomic factors in admissions beyond just GPA.
Dr. Kritsonis has traveled and lectured extensively throughout the United States and world-wide. Some international travels include Australia, New Zealand, Tasmania, Turkey, Italy, Greece, Monte Carlo, England, Holland, Denmark, Sweden, Finland, Russia, Estonia, Poland, Germany, Mexico, the Caribbean Islands, Mexico, Switzerland, Grand Cayman, Haiti, St. Maarten, St. John, St. Thomas, St. Croix, St. Lucia, Puerto Rico, Nassau, Freeport, Jamaica, Barbados, Martinique, Canada, Curacao, Costa Rico, Aruba, Venezuela, Panama, Bora Bora, Tahiti, Latvia, Spain, Honduras, and many more. He has been invited to lecture and serve as a guest professor at many universities across the nation and abroad.
Dr. Kritsonis has traveled and lectured extensively throughout the United States and world-wide. Some international travels include Australia, New Zealand, Tasmania, Turkey, Italy, Greece, Monte Carlo, England, Holland, Denmark, Sweden, Finland, Russia, Estonia, Poland, Germany, Mexico, the Caribbean Islands, Mexico, Switzerland, Grand Cayman, Haiti, St. Maarten, St. John, St. Thomas, St. Croix, St. Lucia, Puerto Rico, Nassau, Freeport, Jamaica, Barbados, Martinique, Canada, Curacao, Costa Rico, Aruba, Venezuela, Panama, Bora Bora, Tahiti, Latvia, Spain, Honduras, and many more. He has been invited to lecture and serve as a guest professor at many universities across the nation and abroad.
IRJET- A Conceptual Framework to Predict Academic Performance of Students usi...IRJET Journal
This document presents a conceptual framework for predicting student academic performance using classification algorithms. The framework uses factors like socioeconomic status, psychological attributes, cognitive attributes, and lifestyle to analyze student performance based on their semester GPA. The document proposes classifying student performance into three classes (first class, second class, third class) based on their first semester GPA. Various classification algorithms like Naive Bayes, random forest, and bagging are evaluated on the student data to identify the best model for predicting performance. The conceptual framework is intended to guide the development of a recommendation system that can help educational institutions identify at-risk students early and improve student outcomes.
Clustering Students of Computer in Terms of Level of ProgrammingEditor IJCATR
Educational data mining (EDM) is one of the applications of data mining. In educational data mining, there are two key domains, i.e. student domain and faculty domain. Different type of research work has been done in both domains.
In existing system the faculty performance has calculated on the basis of two parameters i.e. Student feedback and the result of student in that subject. In existing system we define two approaches one is multiple classifier approach and the other is a single classifier approach and comparing them, for relative evaluation of faculty performance using data mining
Techniques. In multiple classifier approach K-nearest neighbor (KNN) is used in first step and Rule based classification is used in the second step of classification while in single classifier approach only KNN is used in both steps of classification.
But in proposed system, I will analyse the faculty performance using 4 parameters i.e., student complaint about faculty, Student review feedback for faculty, students feedback, and students result etc.
For this proposed system I will be going to use opinion mining technique for analyzing performance of faculty and calculating score of each faculty.
THE USE OF COMPUTER-BASED LEARNING ASSESSMENT FOR PROFESSIONAL COURSES: A STR...IAEME Publication
Background/Objectives: While the increase in classroom technology, it is necessary to examine how assessment is administered through technology. The purpose of this study is to understand how students and faculty are perceived and examine the effectiveness of the computer-based assessment in professional education courses (Educational Technology) at Northern Iloilo Polytechnic State College, Iloilo, Philippines. Methods: The research design utilized in this study is mixed-method research. A computer-based assessment was utilized to assess students' performance in educational technology. This instrument was validated, and pilot tested to establish reliability. Each campus of NIPSC selected ten students of 70 as respondents during Academic Year 2016-2017. Frequency count, mean, standard deviation, and Wilcoxon signed-rank test were statistical tools used for data analyses. Findings: The study's finding showed a high score of students in the posttest ensured better performance of the students in educational technology. The increase in the posttest per performance level of the students was due to an accurate measure of what they have learned in educational technology. The majority of students users agreed that online assessment was fasters than the paper and pencil form. Also, users agreed that online assessment is contemporary and more systematic. They also stated that online assessment is consistent with the teaching style, but they are less anxious. Furthermore, according to faculty and students, ninety percent (90%) believed that computer-based assessment accurately measures what they are teaching and what they learned in school, respectively. Novelty: With the current situation that the education system is in new normal, computer-based learning is important in flexible learning. And assessment using technology is a great help to both faculty and students. Thus, state universities and colleges (SUCs) should adopt this innovation to help teaching and learning.
“E-SurveyCo MIT: A Electronic Survey of Community Perception, Awareness, Acce...CatleenMadayag
An educational institution's guiding principles are its vision, mission, goals, and objectives. The different
accreditation body such as Accrediting Agency of Chartered Colleges and Universities in the Philippines,
Inc. (AACCUP), believe that the degree to which VMGOs are attained is used to evaluate a university. Thus
this study was conducted to measure the stakeholder's perceptions, awareness, acceptance, and
understanding towards the Isabela State University VMGO. Lack of in-person access to participants and
timeliness have pushed researcher to use electronic surveys to connect to the stakeholders of the university
due to COVID-19 pandemic. The stakeholders of the University are the respondents of the study and it was
categorized into two namely the Internal and External stakeholders. Internal stakeholders are composed of
Administrators or faculty members, non-teaching staff, and students while External stakeholders are
Parents/guardians, alumni, and industries, linkages, or cooperating agencies. A descriptive research
methodology was used by the researcher. Data from the respondents were gathered through convenience
sampling. Result showed that majority of the respondents were very highly aware of the University vision,
mission, college goals and objectives and highly aware on the dissemination activities by the university.
They also had very high acceptance and understanding of the university vision, mission, and campus goals.
The stakeholders also satisfy to the clarity and consistency of the VMGO. In terms of the program
objectives, the respondents were very highly aware of their specific programs and they have a very high
acceptance on their specific program objectives. Study also shows that stakeholder’s perception the VMGO
of the university is congruent with activities, practices, projects, and operations. Based on the foregoing, it
is recommended that Isabela State University should continuously work for the awareness and acceptance
of the ISU VMGO. It is recommended that both the College and CGS make more efforts in disseminating
the goals and objective of the MIT program especially to the parents. Activities must be more focused to
make them feel that the activities of the school is geared towards the realization of the VMGO. The
researcher also proofs that using online survey methods can be particularly useful and successful during
the current pandemic thus the researcher encourages others to utilized also e-survey to measure the
Perception, Awareness, Acceptance and Understanding of the vision, mission, goals of the college and
objective of the program.
Accurate prediction and early identification of student at-risk of attrition are of high concern for higher
educational institutions (HEIs). It is of a great importance not only to the students but also to the
educational administrators and the institutions in the areas of improving academic quality and
efficient utilisation of the available resources for effective intervention. However, despite the different
frameworks and various models that researchers have used across institutions for predicting performance,
only negligible success has been recorded in terms of accuracy, efficiency and reduction of student
attrition. This has been attributed to the inadequate and selective use of variables for the predictive models.
AN INTEGRATED SYSTEM FRAMEWORK FOR PREDICTING STUDENTS’ ACADEMIC PERFORMANCE ...ijcsit
Accurate prediction and early identification of student at-risk of attrition are of high concern for higher educational institutions (HEIs). It is of a great importance not only to the students but also to the educational administrators and the institutions in the areas of improving academic quality and efficient utilisation of the available resources for effective intervention. However, despite the different frameworks and various models that researchers have used across institutions for predicting performance, only negligible success has been recorded in terms of accuracy, efficiency and reduction of student
attrition. This has been attributed to the inadequate and selective use of variables for the predictive models. This paper presents a multi-dimensional and an integrated system framework that involves considerable learners’ input and engagement in predicting their academic performance and intervention in HEIs. The purpose and functionality of the framework are to produce a comprehensive, unbiased and efficient way of predicting student performance that its implementation is based upon multi-sources data and database
system. It makes use of student demographic and learning management system (LMS) data from the institutional databases as well as the student psychosocial-personality (SPP) data from the survey collected from the student to predict performance. The proposed approach will be robust, generalizable, and possibly give a prediction at a higher level of accuracy that educational administrators can rely on for providing timely intervention to students.
Accurate prediction and early identification of student at-risk of attrition are of high concern for higher educational institutions (HEIs). It is of a great importance not only to the students but also to the educational administrators and the institutions in the areas of improving academic quality and efficient utilisation of the available resources for effective intervention. However, despite the different frameworks and various models that researchers have used across institutions for predicting performance, only negligible success has been recorded in terms of accuracy, efficiency and reduction of student attrition. This has been attributed to the inadequate and selective use of variables for the predictive models. This paper presents a multi-dimensional and an integrated system framework that involves considerable learners’ input and engagement in predicting their academic performance and intervention in HEIs. The purpose and functionality of the framework are to produce a comprehensive, unbiased and efficient way of predicting student performance that its implementation is based upon multi-sources data and database system. It makes use of student demographic and learning management system (LMS) data from the institutional databases as well as the student psychosocial-personality (SPP) data from the survey collected from the student to predict performance. The proposed approach will be robust, generalizable, and possibly give a prediction at a higher level of accuracy that educational administrators can rely on for providing timely intervention to students. --
Fuzzy Measurement of University Students Importance Indexes by Using Analytic...IRJESJOURNAL
Purpose: The purpose of this paper is to apply a Fuzzy Analytical Hierarchy Process model FAHP for estimating students' importance indexes problem, where the measures of students' attitudes and responses are often uncertain or difficult to determine by using non-fuzzy model. Fuzzy set theory treats a kind of uncertainty called fuzziness. It shows that the boundary of “yes” or “no” is ambiguous and appears in the meaning of words or included in the subjunctives or recognition of human beings. Design/methodology/approach: This paper adapts FAHP to analysis students' satisfaction for the services submitted by the university to enhance learning process and circumstance environment for students, this area of researches is still out of university importance. For the purposes of the survey, questionnaires were designed for all the factors which are affecting in students' satisfaction and most probably all these factors are taken according to their suitable priority. There were five main criterions in the questionnaires. Criterion one focuses on administrative university services Si, the second criterion concentrate with teaching and learning process Ti, while the third criterion is university climate and student activities Ai, the fourth criterion covers the administrative facilities Fi, and the fifth criterion is learning facilities students Li, were asked to measure their perceived experiences with those criterions. Each criterion contains some of sub criteria. Findings: The results illustrate that the quality of teachers T1, Opportunities for recreational activities A1, fair evaluation for students T6, and remedial support A3 are the most important indexes for students. The proposed model would help decision-makers to enhancement the quality of the services and may be adding other facility to make the university more attractive. Research limitations/implications: This paper makes some assumptions such as the number of sample are taken from seven colleges only and not cover all colleges in the university and concentrated about the final levels. Originality/value: This paper introduces fuzzy theory with AHP approach to the research of university services as customers in public firms and it has reached some valuable conclusions, which has opened up a new field of study in the universities services area
The study sought to establish the impact of electronic information services (EIS) on students and alumni of Glasgow Caledonian University. Questionnaires were distributed to current students and alumni to collect both quantitative and qualitative data on EIS usage and information literacy. The data was analyzed to compare results between groups and identify commonly used resources. It was found that respondents understood the concept of information literacy, which alumni felt gave them an advantage in their careers. Unemployed alumni were less engaged with information literacy skills. The study concluded that information literacy skills learned in university could be directly applied and useful in the workplace.
Predicting students’ intention to continue business courses on online platfor...Samsul Alam
The objective of this study was to analyze the intention of a University's business department students to continue their studies on e-learning platforms during the ongoing COVID-19 pandemic. To this end, a questionnaire was developed to collect primary data from students in business fields. The study took into account more than 285 respondents from two different universities and relied on the expectation confirmation model (ECM) theory and the structural equation model. The partial least squares (SEM-PLS) method was used to analyze the data. The results of the study showed that task skills (TS) and task challenges (TC) were significant for the enjoyment (EN) of the students which in turn had a positive effect on the satisfaction levels. Confirmation (CON) had an impact on the post adoption perceived usefulness (PAPU), which was deemed positive for student satisfaction (SAT). The SAT and psychological safety (PS) of online learning platforms were found to positively influence the continuance intention (CI) on e-learning platforms. Finally, both SAT and PS of online learning platforms were observed to positively influence CI on e-learning platforms. Further research in this area could be useful in making decisions about promoting educational programs based on e-learning. The researchers recommend that academicians and policymakers must ensure appropriate arrangements for teaching on e-learning platforms.
PREDICTING ACADEMIC MAJOR OF STUDENTS USING BAYESIAN NETWORKS TO THE CASE OF ...ijcax
In this study, which took place current year in the city of Maragheh in IRAN. Number of high school students in the fields of study: mathematics, Experimental Sciences, humanities, vocational, business and science were studied and compared. The purpose of this research is to predict the academic major of high school students using Bayesian networks. The effective factors have been used in academic major selection
for the first time as an effective indicator of Bayesian networks. Evaluation of Impacts of indicators on each other, discretization data and processing them was performed by GeNIe. The proper course would be advised for students to continue their education.
PREDICTING ACADEMIC MAJOR OF STUDENTS USING BAYESIAN NETWORKS TO THE CASE OF ...ijcax
In this study, which took place current year in the city of Maragheh in IRAN. Number of high school
students in the fields of study: mathematics, Experimental Sciences, humanities, vocational, business and
science were studied and compared. The purpose of this research is to predict the academic major of high
school students using Bayesian networks. The effective factors have been used in academic major selection
for the first time as an effective indicator of Bayesian networks. Evaluation of Impacts of indicators on
each other, discretization data and processing them was performed by GeNIe. The proper course would be
advised for students to continue their education.
PREDICTING ACADEMIC MAJOR OF STUDENTS USING BAYESIAN NETWORKS TO THE CASE OF ...ijcax
In this study, which took place current year in the city of Maragheh in IRAN. Number of high school students in the fields of study: mathematics, Experimental Sciences, humanities, vocational, business and
science were studied and compared. The purpose of this research is to predict the academic major of high school students using Bayesian networks. The effective factors have been used in academic major selection for the first time as an effective indicator of Bayesian networks. Evaluation of Impacts of indicators on each other, discretization data and processing them was performed by GeNIe. The proper course would be
advised for students to continue their education.
Predicting student performance in higher education using multi-regression modelsTELKOMNIKA JOURNAL
Supporting the goal of higher education to produce graduation who will be a professional leader is a crucial. Most of universities implement intelligent information system (IIS) to support in achieving their vision and mission. One of the features of IIS is student performance prediction. By implementing data mining model in IIS, this feature could precisely predict the student’ grade for their enrolled subjects. Moreover, it can recognize at-risk students and allow top educational management to take educative interventions in order to succeed academically. In this research, multi-regression model was proposed to build model for every student. In our model, learning management system (LMS) activity logs were computed. Based on the testing result on big students datasets, courses, and activities indicates that these models could improve the accuracy of prediction model by over 15%.
Performance of student is dependent on their subject select and faculty’s expertise who teach the subject. Sometimes subject selection done by the student. Few students’ select subject wisely, few of them select because of friend influence, faculty influence, without thinking on it. To reduce the cost and related overhead department tries to reduce the variety of elective. Wisely selected and offered subject will increase the performance of student and throughput of the faculty. In this paper a recommender system is proposed which find the score of knowledge level for student and score of faculty score. These scores will be used to recommend the right subject to adept faculty of that subject.
A Methodology To Analyze Self-Reflection In E-PortfoliosRick Vogel
This document presents a methodology for analyzing self-reflection in e-portfolios created by students in the Grand Challenge Scholars Program at the University of Maryland, Baltimore County. The methodology applies two existing frameworks - Robert Grossman's levels of reflection and AAC&U's Integrative Learning rubric - to assess self-reflection. An initial analysis of e-portfolios from the first two cohorts found students did not reach high levels of reflection with a generic prompt. The analysis is ongoing and expects to see deeper reflection in newer portfolios with changes to expectations and prompts. The goal is to evaluate the program and identify interventions to enhance students' learning experiences.
At Digidev, we are working to be the leader in interactive streaming platforms of choice by smart device users worldwide.
Our goal is to become the ultimate distribution service of entertainment content. The Digidev application will offer the next generation television highway for users to discover and engage in a variety of content. While also providing a fresh and
innovative approach towards advertainment with vast revenue opportunities. Designed and developed by Joe Q. Bretz
The Evolution of the Leonardo DiCaprio Haircut: A Journey Through Style and C...greendigital
Leonardo DiCaprio, a name synonymous with Hollywood stardom and acting excellence. has captivated audiences for decades with his talent and charisma. But, the Leonardo DiCaprio haircut is one aspect of his public persona that has garnered attention. From his early days as a teenage heartthrob to his current status as a seasoned actor and environmental activist. DiCaprio's hairstyles have evolved. reflecting both his personal growth and the changing trends in fashion. This article delves into the many phases of the Leonardo DiCaprio haircut. exploring its significance and impact on pop culture.
Top IPTV UK Providers of A Comprehensive Review.pdfXtreame HDTV
The television landscape in the UK has evolved significantly with the rise of Internet Protocol Television (IPTV). IPTV offers a modern alternative to traditional cable and satellite TV, allowing viewers to stream live TV, on-demand videos, and other multimedia content directly to their devices over the internet. This review provides an in-depth look at the top IPTV UK providers, their features, pricing, and what sets them apart.
Unveiling Paul Haggis Shaping Cinema Through Diversity. .pdfkenid14983
Paul Haggis is undoubtedly a visionary filmmaker whose work has not only shaped cinema but has also pushed boundaries when it comes to diversity and representation within the industry. From his thought-provoking scripts to his engaging directorial style, Haggis has become a prominent figure in the world of film.
Modern Radio Frequency Access Control Systems: The Key to Efficiency and SafetyAITIX LLC
Today's fast-paced environment worries companies of all sizes about efficiency and security. Businesses are constantly looking for new and better solutions to solve their problems, whether it's data security or facility access. RFID for access control technologies have revolutionized this.
Leonardo DiCaprio House: A Journey Through His Extravagant Real Estate Portfoliogreendigital
Introduction
Leonardo DiCaprio, A name synonymous with Hollywood excellence. is not only known for his stellar acting career but also for his impressive real estate investments. The "Leonardo DiCaprio house" is a topic that piques the interest of many. as the Oscar-winning actor has amassed a diverse portfolio of luxurious properties. DiCaprio's homes reflect his varied tastes and commitment to sustainability. from retreats to historic mansions. This article will delve into the fascinating world of Leonardo DiCaprio's real estate. Exploring the details of his most notable residences. and the unique aspects that make them stand out.
Follow us on: Pinterest
Leonardo DiCaprio House: Malibu Beachfront Retreat
A Prime Location
His Malibu beachfront house is one of the most famous properties in Leonardo DiCaprio's real estate portfolio. Situated in the exclusive Carbon Beach. also known as "Billionaire's Beach," this property boasts stunning ocean views and private beach access. The "Leonardo DiCaprio house" in Malibu is a testament to the actor's love for the sea and his penchant for luxurious living.
Architectural Highlights
The Malibu house features a modern design with clean lines, large windows. and open spaces blending indoor and outdoor living. The expansive deck and patio areas provide ample space for entertaining guests or enjoying a quiet sunset. The house has state-of-the-art amenities. including a gourmet kitchen, a home theatre, and many guest suites.
Sustainable Features
Leonardo DiCaprio is a well-known environmental activist. whose Malibu house reflects his commitment to sustainability. The property incorporates solar panels, energy-efficient appliances, and sustainable building materials. The landscaping around the house is also designed to be water-efficient. featuring drought-resistant plants and intelligent irrigation systems.
Leonardo DiCaprio House: Hollywood Hills Hideaway
Privacy and Seclusion
Another remarkable property in Leonardo DiCaprio's collection is his Hollywood Hills house. This secluded retreat offers privacy and tranquility. making it an ideal escape from the hustle and bustle of Los Angeles. The "Leonardo DiCaprio house" in Hollywood Hills nestled among lush greenery. and offers panoramic views of the city and surrounding landscapes.
Design and Amenities
The Hollywood Hills house is a mid-century modern gem characterized by its sleek design and floor-to-ceiling windows. The open-concept living space is perfect for entertaining. while the cozy bedrooms provide a comfortable retreat. The property also features a swimming pool, and outdoor dining area. and a spacious deck that overlooks the cityscape.
Environmental Initiatives
The Hollywood Hills house incorporates several green features that are in line with DiCaprio's environmental values. The home has solar panels, energy-efficient lighting, and a rainwater harvesting system. Additionally, the landscaping designed to support local wildlife and promote
Everything You Need to Know About IPTV Ireland.pdfXtreame HDTV
The way we consume television has evolved dramatically over the past decade. Internet Protocol Television (IPTV) has emerged as a popular alternative to traditional cable and satellite TV, offering a wide range of channels and on-demand content via the internet. In Ireland, IPTV is rapidly gaining traction, with Xtreame HDTV being one of the prominent providers in the market. This comprehensive guide will delve into everything you need to know about IPTV Ireland, focusing on Xtreame HDTV, its features, benefits, and how it is revolutionizing TV viewing for Irish audiences.
Orpah Winfrey Dwayne Johnson: Titans of Influence and Inspirationgreendigital
Introduction
In the realm of entertainment, few names resonate as Orpah Winfrey Dwayne Johnson. Both figures have carved unique paths in the industry. achieving unparalleled success and becoming iconic symbols of perseverance, resilience, and inspiration. This article delves into the lives, careers. and enduring legacies of Orpah Winfrey Dwayne Johnson. exploring how their journeys intersect and what we can learn from their remarkable stories.
Follow us on: Pinterest
Early Life and Backgrounds
Orpah Winfrey: From Humble Beginnings to Media Mogul
Orpah Winfrey, often known as Oprah due to a misspelling on her birth certificate. was born on January 29, 1954, in Kosciusko, Mississippi. Raised in poverty by her grandmother, Winfrey's early life was marked by hardship and adversity. Despite these challenges. she demonstrated a keen intellect and an early talent for public speaking.
Winfrey's journey to success began with a scholarship to Tennessee State University. where she studied communication. Her first job in media was as a co-anchor for the local evening news in Nashville. This role paved the way for her eventual transition to talk show hosting. where she found her true calling.
Dwayne Johnson: From Wrestling Royalty to Hollywood Superstar
Dwayne Johnson, also known by his ring name "The Rock," was born on May 2, 1972, in Hayward, California. He comes from a family of professional wrestlers, with both his father, Rocky Johnson. and his grandfather, Peter Maivia, being notable figures in the wrestling world. Johnson's early life was spent moving between New Zealand and the United States. experiencing a variety of cultural influences.
Before entering the world of professional wrestling. Johnson had aspirations of becoming a professional football player. He played college football at the University of Miami. where he was part of a national championship team. But, injuries curtailed his football career, leading him to follow in his family's footsteps and enter the wrestling ring.
Career Milestones
Orpah Winfrey: The Queen of All Media
Winfrey's career breakthrough came in 1986 when she launched "The Oprah Winfrey Show." The show became a cultural phenomenon. drawing millions of viewers daily and earning many awards. Winfrey's empathetic and candid interviewing style resonated with audiences. helping her tackle diverse and often challenging topics.
Beyond her talk show, Winfrey expanded her empire to include the creation of Harpo Productions. a multimedia production company. She also launched "O, The Oprah Magazine" and OWN: Oprah Winfrey Network, further solidifying her status as a media mogul.
Dwayne Johnson: From The Ring to The Big Screen
Dwayne Johnson's wrestling career took off in the late 1990s. when he became one of the most charismatic and popular figures in WWE. His larger-than-life persona and catchphrases endeared him to fans. making him a household name. But, Johnson had ambitions beyond the wrestling ring.
In the early 20
Christian Louboutin: Innovating with Red Solesget joys
Christian Louboutin is celebrated for his innovative approach to footwear design, marked by his trademark red soles. This in-depth look at his life and career explores the origins of his creativity, the milestones in his journey, and the impact of his work on the fashion industry. Learn how Louboutin's bold vision and dedication to excellence have made his brand synonymous with luxury and style.
The Unbelievable Tale of Dwayne Johnson Kidnapping: A Riveting Sagagreendigital
Introduction
The notion of Dwayne Johnson kidnapping seems straight out of a Hollywood thriller. Dwayne "The Rock" Johnson, known for his larger-than-life persona, immense popularity. and action-packed filmography, is the last person anyone would envision being a victim of kidnapping. Yet, the bizarre and riveting tale of such an incident, filled with twists and turns. has captured the imagination of many. In this article, we delve into the intricate details of this astonishing event. exploring every aspect, from the dramatic rescue operation to the aftermath and the lessons learned.
Follow us on: Pinterest
The Origins of the Dwayne Johnson Kidnapping Saga
Dwayne Johnson: A Brief Background
Before discussing the specifics of the kidnapping. it is crucial to understand who Dwayne Johnson is and why his kidnapping would be so significant. Born May 2, 1972, Dwayne Douglas Johnson is an American actor, producer, businessman. and former professional wrestler. Known by his ring name, "The Rock," he gained fame in the World Wrestling Federation (WWF, now WWE) before transitioning to a successful career in Hollywood.
Johnson's filmography includes blockbuster hits such as "The Fast and the Furious" series, "Jumanji," "Moana," and "San Andreas." His charismatic personality, impressive physique. and action-star status have made him a beloved figure worldwide. Thus, the news of his kidnapping would send shockwaves across the globe.
Setting the Scene: The Day of the Kidnapping
The incident of Dwayne Johnson's kidnapping began on an ordinary day. Johnson was filming his latest high-octane action film set to break box office records. The location was a remote yet scenic area. chosen for its rugged terrain and breathtaking vistas. perfect for the film's climactic scenes.
But, beneath the veneer of normalcy, a sinister plot was unfolding. Unbeknownst to Johnson and his team, a group of criminals had planned his abduction. hoping to leverage his celebrity status for a hefty ransom. The stage was set for an event that would soon dominate worldwide headlines and social media feeds.
The Abduction: Unfolding the Dwayne Johnson Kidnapping
The Moment of Capture
On the day of the kidnapping, everything seemed to be proceeding as usual on set. Johnson and his co-stars and crew were engrossed in shooting a particularly demanding scene. As the day wore on, the production team took a short break. providing the kidnappers with the perfect opportunity to strike.
The abduction was executed with military precision. A group of masked men, armed and organized, infiltrated the set. They created chaos, taking advantage of the confusion to isolate Johnson. Johnson was outnumbered and caught off guard despite his formidable strength and fighting skills. The kidnappers overpowered him, bundled him into a waiting vehicle. and sped away, leaving everyone on set in a state of shock and disbelief.
The Immediate Aftermath
The immediate aftermath of the Dwayne Johnson kidnappin
From Teacher to OnlyFans: Brianna Coppage's Story at 28get joys
At 28, Brianna Coppage left her teaching career to become an OnlyFans content creator. This bold move into digital entrepreneurship allowed her to harness her creativity and build a new identity. Brianna's experience highlights the intersection of technology and personal branding in today's economy.
Barbie Movie Review - The Astras.pdffffftheastras43
Barbie Movie Review has gotten brilliant surveys for its fun and creative story. Coordinated by Greta Gerwig, it stars Margot Robbie as Barbie and Ryan Gosling as Insight. Critics adore its perky humor, dynamic visuals, and intelligent take on the notorious doll's world. It's lauded for being engaging for both kids and grown-ups. The Astras profoundly prescribes observing the Barbie Review for a delightful and colorful cinematic involvement.https://theastras.com/hca-member-gradebooks/hca-gradebook-barbie/
ESTIMATION OF PERSISTENCE AT A COMMUNITY COLLEGE: A COMPARISON OF ALTERNATIVE MACHINE LEARNING MODELS
1. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
DOI : 10.5121/ijcsea.2023.13101 1
ESTIMATION OF PERSISTENCE AT A COMMUNITY
COLLEGE: A COMPARISON OF ALTERNATIVE
MACHINE LEARNING MODELS
Fermin Ornelas
Institutional Research, Rio Salado College, Maricopa Community Colleges,
Tempe, AZ, USA.
ABSTRACT
This research focuses on developing persistence models for Rio Salado College. It is an initial effort to
predict persistence from one term to the next. Several ensemble models are experimented and compared in
their respective key metrics such as: confusion matrix, AUC, F1-Score, and feature importance.
Exploratory data analysis is undertaken to narrow the set of variables utilized in the models. Two models
were considered for possible implementation: a logistic regression and a gradient boosting machine. The
former is easier to implement and explain to non-technical personnel, while the latter behaves like a black
box. Based on key performance metrics, the model of choice was the gradient boosting machine.
Development and testing were conducted with python using jupyter notebooks. The author hopes that this
experimental process will fill a vacuum in the analytical needs of community colleges.
KEYWORDS
accuracy, AUC, confusion matrix, ensemble models, feature importance, gradient boosting machines,
Machine Learning, persistence, precision, recall
1. INTRODUCTION
In the current economic conditions, academic institutions continue making efforts to adapt to the
student needs. The pandemic surge followed by high levels of unemployment has made remote
working and learning necessary alternatives for both workers [2], [3] and students to continue
providing for their families and enhancing their working and academic skills, respectively.
In previous recessions, higher education institutions experienced countercyclical enrolment
increases, but during the current COV19 driven recession that has not been the case. Higher
education is currently having to reduce personnel or put on hold additional hiring and is suffering
from enrolment declines due to the pandemic and degrading economic conditions, [15].
According to [22], overall enrolment is down by 3% with freshman enrolment being down to
13%. These declines are more severe for 2-year colleges than for 4-year institutions. Community
colleges according to [23] data are experiencing a 19% decline with a pronounced decline among
students of colour. Moreover, a recent survey conducted by an education marketing company
[14], among 528 students across the U.S. concluded that 43% of prospective students for one and
2-year programs are considering delaying enrolment. Other findings of the survey were: 54% of
traditional age students considered having a degree or certificate extremely valuable; 65% of non-
traditional students because of the pandemic thought that a degree or certificate was extremely
valuable. Considering these facts, by necessity educational institutions are increasingly creative
in trying to retain and support their enrolled students. The prolonged pandemic and the sudden
2. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
2
recession have caught everyone by surprise and some not well prepared for working remotely
while providing training and education to a smaller student population in an online environment.
After the Great Recession, all institutions, experienced dynamic student growth reaching a peak
around 2010. Over 8 million students enrolled at 2-year colleges [16]. Since then, by 2017,
enrolment at these institutions has declined by about 1 million [5]. It is this scenario that has all
institutions and Rio in particular, focusing their efforts on retaining current students. Henceforth,
Rio Salado College has been innovative in finding ways to leverage predictive analytics to
enhance academic success and persistence among its diverse student population. It has developed
a course success tracking system, Rio PACE, to provide timely feedback to students and
instructors. In addition to its course success predictive modelling strategies [24], [27], the
researcher is currently working on another modelling project on persistence. The objectives are to
understand student’s behaviour and leverage empirical results to foster persistence and retention
while leveraging Rio Salado College resources to effectively support its student population.
Persistence, a dichotomous event, for this project is defined as enrolling in a term (Fall 2021) and
continuing into the next term (Spring 2022). However, [23] defines persistence as continuing
studying at any institution for a second year. In addition, the report concludes that 6 in 10
students persist into the second fall term. The definition to be used here obeys two reasons: a
concern for students’ continuity in their studies and the fact that the majority of Rio College
student population are not pursuing an associate degree or certificate. That has created two types
of students: those attending to take specific courses often to transfer the credits to a major
university; and a two-prong segment pursuing 1-year certificates or associate degrees. Thus, this
investigation will seek to identify features that motivate non-degree, degree, and certificate
seekers to persist or not into the next term.
There is also a technical complexity associated with development, implementation and tracking
of machine learning models. Anxious in their quest to serve students better, colleges are
contracting with vendors to effectively attend to their student population needs with timely advice
and to prevent further enrolment declines. Those vendors, often suggest complex solution
packages that 2-year colleges cannot afford or may not fit their analytical needs. That has resulted
in some discomfort on the use of such applications and subsequent recommendations [6], [13],
[29]. These solutions behave like black boxes understood only by vendors’ technical personnel.
In a recent paper [4], it is reported that researchers and college administrators have little practical
means of assessing predictive software results potentially having adverse effects for students.
Furthermore, we intend to compare analytical solutions for various ensemble classifiers: logistic
regression, decision trees, gradient boosting, random forest, ADABOOST, and XGBOOST. The
research will be developed using easily available software tools and LMS inhouse data, hoping to
contribute to the analytics needs of 2-year college institutions.
Therefore, the current investigation project has the following objectives: to develop persistence a
model for two populations, degree and certificate, and non-degree seeking students; to identify
features that drive student persistence; and to compare results from all the different models.
Therefore, the article will discuss literature review, followed by methodological aspects of the
model, next it will present empirical findings, followed by limitations of the study. Finally, it will
render conclusions and recommendation arising from this research.
3. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
3
2. LITERATURE REVIEW
Predicting persistence at community colleges has always been a challenging issue. Some studies
have attempted to predict persistence using pre-enrolment high school variables and focus on first
time in college students. Most have attempted to predict persistence focusing on a sense of
belonging and involvement at 4-year institutions, [18], [30]. This research focuses on persistence
research and estimation issues related to community colleges using machine learning modelling.
In [4] a study to evaluate whether students at the Virginia Community College system graduated
with a college level credential within six years of initial entry. They compared alternative models
on two dimensions: random truncation of a current cohort sample aligned to enrolment length
distribution of historical cohorts, and different variable construction, i.e., term specific and
constructed. The pursued models were, OLS regression, Cox proportional hazard, random forest,
and XGBoost. Over 331 predicting features fitted into the models utilized to predict graduation.
[4] concluded the models predicted reasonably well. However, the predictions differed among
models in the students’ probability ranking, depending on the alternative variables tried. They
expressed concerns that this variability could result in inefficient use of resources for targeting
students and potential bias against underserved students. [4] observed that no substantial
increases in accuracy existed when applying more complex models. In another interesting
research [28], as part of their modelling task to predict MOOCs dropouts’ online behaviour, built
predictive features based on their expertise and crowd sourcing. Some in the latter group were
provided an artificial data set and encouraged to write their own scripts for feature engineering.
This approach resulted in a richer set of features.
The self-proposed approach rendered 18 features, while crowd sourcing resulted in another 10
new covariates. Some of these variables had high predicting power. These features were tested on
students enrolled in a Circuit and Electronics course during Fall 2012. Randomized logit was the
selected model and students were divided into four cohorts: passive collaborator, wiki
contributor, forum contributor, and fully collaborators. Predictive power of the engineered
features varied by cohort.
In [25], an investigation to test a comprehensive list of factors impacting student retention at the
institutional level. Employing a logit and a random forest model they discovered that GPA,
institution’s primary campus, first generation in college, and academic advisor were the primary
factors impacting student retention behaviour. The random forest model had, in descending order,
primary campus location, first generation, age, GPA, and academic advisor as the top five
features important predictors of student retention. Overall model accuracy was reported at 78%,
no other metric was provided. The logit model revealed that academic advisor, support services,
primary campus location, and academic achievements were relevant for student retention.
Focusing on 9,200 first time community college students (FTIC) over a four-year period, [8],
studied fall-to-spring and fall-to-fall retention. The time window started from Fall 2001 to Fall
2004. They found that developmental education and online courses had impact on student
persistence. Other predictors were, financial aid, parent’s education, ethnicity, hours enrolled and
dropped per term in the first fall semester, and support services participation. Stronger predictors
for fall-to-spring retention, in descending order, were: passing developmental reading, taking an
online course, participating in student services, not taking a developmental reading course,
passing a developmental math course, receiving financial aid, father with some college, hours
enrolled in the first fall term, and student age. Unfavourable odds of retention were: not taking
developmental math, mother’s education, and hours dropped in the first fall term. Similarly, for
fall-to-fall retention, about the same variables were found to positively impact retention, except
4. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
4
for passing a developmental writing course and mother with college education. Age was found to
be non-significant in predicting retention.
In another early study, [7], identified predictors of attrition before community college students
began classes by focusing on pre-enrolment factors. Using a logistic classifier, main findings
were: high school GPA was the strongest predictor, the lower the GPA the greater the odds of
dropping out, age indicator 20-24, being minority except for Asian and enrolled part-time.
Interestingly, the researcher found no impact for remedial course in reading, math, and writing.
Tracking revealed that students planning to graduate, or transfer had higher retention rates. A
caveat on models with ethnicity variables, minimizing bias would require further refinement
when predicting non-retention of minorities [4], [6].
In a retention study at CSU - San Bernardino, [15] focused on yearly retention from freshmen to
sophomore year for first time freshmen entering in the fall 2009 and fall 2010 quarters. They
wanted to identify predictors of retention and the likelihood of students dropping out. The final
logistic classifier included the following features: ethnicity, high school GPA, university studies
100 enrolment, first term GPA, percent of courses completed in the first year, and number of
GENED course taken in the first year. Decile probability scores were bucketed into high,
medium, and low risk of not being retained. At-risk students would be encouraged by peer
advisors to engage in university activities such as: service and community learning,
diversity/global experiences, learning communities and undergraduate research.
At NYIT, [2], report on an end-to-end model implementation to predict first-year retention. The
data collected for this process came from admission applications, registration/placement testing,
surveys on students when taking the Compass placement test, and financial information on 1,453
students during the fall of 2011 and 2012. Based on this sample a year later only 983 returned,
68% retention rate. The models tried were logistic regression, neural networks, naïve Bayesian,
decision tree, and ensemble. The latter was the model of choice with recall values of 73% and
precision of 54% in the validation data set. Furthermore, the prediction results were shared with
counselling staff who would use the report for at-risk student treatment.
To improve retention and academic achievement at a community college, [17], conducted a
treatment and control experiment where students in the former group attended a freshmen
seminar taught by specially trained instructors/advisors while the control group were taught by
conventional instructors. Results proved that those in the treatment group achieved higher GPA
and were retained into the immediate term at a higher rate than those in the control group. The
experiment was applied to 280 students, 158 and 122 were in the treatment and control group,
respectively. The 14 sections for the course were limited to traditional sections only during the
Fall 2009 and Spring 2010.
In another study at a community college in Southeast Texas, [12] evaluated the impact of a
student success course (SSC) on three metrics: persistence, retention, and academic achievement.
Employing the CCSSE survey for items 4 and 9 in a sample of 432 students, 197, and 235 in the
treatment and control groups, respectively. [12] conducted Chi-square testing for independence
concluding that students enrolled in the SCC program had higher persistence, and retention rates,
and better academic achievement in gatekeeping courses. No attempt was made to build a
predictive model, however that does not diminish the relevance of the study nor the impact of the
SSC on the three outcome measures.
At the University of South Florida, [19], utilizing pre-matriculation characteristics and data
derived from the College Student Expectations Questionnaire built a logistic classifier for
5. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
5
retention under two alternatives: one based on the former type of data; and another one using
both pre-matriculation and survey data. The sample population was incoming FTIC, 3998
students in the Fall and Summer of 2006. The retention rate for Fall 2007 was reported at 82.2%
for the model version with pre-matriculation data and 78.2% for the model using both pre-
matriculation and survey data. Correctly predicted observations were reported as 82.3% and
78.9% under each alternative model, respectively.
For the first model, variables having a positive effect on persistence were as follows: high school
GPA, Asian vs White, Black vs. White, and time elapsed since orientation. Adverse role was
reported for SAT combined, pre-nursing major, and residence status commuter. For the second
model, the factors positively predicting persistence were, high school GPA, being black vs. being
white, expecting to participate in clubs and student organizations, expecting to read textbooks or
assigned books in college, expecting to work on campus. Negatively influencing persistence
were, expecting to read non-assigned books, and expecting to work off-campus while in college.
[19], followed this work with some suggestions for creating interventions to treat at risk of not-
persisting students.
In another persistence study at community colleges [21] intended to explore likely factors
impacting students’ decision whether to drop or stay in school. The authors found that cumulative
GPA was the strongest predictor of persistence followed by units taken and English proficiency.
Other attributes such as age, work hours, and financial aid were initially statistically significant,
but their performance diluted once other variables were entered into the predictive model. This
phenomenon was also observed for psychosocial and academic integration variables.
3. ENSEMBLE MODELS AND METRICS
The models developed for this research belong to supervised learning. The main characteristic of
these classifiers is the existence of a target outcome, in our case, persistence. While logistic
regression is based on the sigmoid function, the other models are derived from decision trees. The
latter are referred as nonlinear nonparametric classifiers because they do not have any
distributional assumptions. They can either tune their learning algorithm or the dataset used to
train them, to ensure diversity and high model performance [11].
Simple models are more likely to introduce bias as they are unable to fit the data because of
underfitting, while complex models are prone to introduce more variance by overfitting the data.
Ensemble models are intended to address both bias and variance by combining predictions of
multiple base models. Models such as bagging, ADABOOST, random forest, gradient boosting,
and XGBOOST employ weak learning and majority voting to arrive at a final model selection
often using decision trees as base models. They are considered generative because they can
generate and affect the base learners they use, [11]. Weak learning is a sequential process where
the errors of the weak predictions are assigned a higher weight compared to correctly classified
predictions when estimating the base models. The errors from these weak learners are used in
subsequent estimation. The objective is to improve the overall predictions. Majority voting or
average voting requires comparison of results from each base classifier to make a final selection,
based on a majority agreement on the outcome of interest prediction. In addition to enhance the
final classifier results often a researcher applies hyperparameter tuning. These are employed
either using grid search or random search techniques to fine tune the final model.
The machine learning model results discussed in this research are compared in their performance
metrics. The confusion matrix is often a 2x2 contingency table representing both correct and
misclassified predictions. True positive and true negatives are identified along the diagonal, while
false positives and false negatives are found in the off-diagonal cells. These values are utilized to
6. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
6
compute the following metrics: recall or true positive rate, specificity or true negative rate, false
positive rate, and precision.
A classification report built by Python sklearn metrics application summarizes these metrics by
outcome class. Another important metric is the ROC curve, it is a plot of the true positive rate
against the false positive rate at different cut-off points for the classifier. Commonly, a threshold
of .5 is established to classify a prediction outcome as either success or failure. In our case, a
probability value above that threshold would mean a student has persisted, otherwise he/she has
not. Note that this value can be changed but it will also affect the prediction classification rates.
There is also another measure that eases model interpretability, feature importance. It gives a
percentage ranking of the features selected by the model and can be used to illustrate users the
impact that those features have on the outcome score. Therefore, this research will demonstrate
the use of these measures to arrive at a final model recommendation. Since model interpretability
is a concern among operational personnel, there are new applications intended to address it. New
packages on such as LIME or SHAP are intended to look deeper in the prediction results and the
modelling features contributing to the predicted probability helping to erase the notion that
machine learning models behave like a black box [20]. These techniques will not be utilized in
this research.
4. EXPLORATORY DATA ANALYSIS
The sample data for model development and validation was extracted from the inhouse MLS
using MSQL queries. The total cleaned sample encompassed 19,459 records for the Fall 2021
and the next term was Spring 2022. Of these, 80% was utilized for model development and 20%
remained for validation, 15,567 and 3,892 observations, respectively. The actual persistence rate
for this data is reported at 53% and 47% as non-persisting. These figures avoid issues of
unbalanced data and ease the estimation process. As part of the exploratory data analysis (EDA),
descriptive statistics were calculated for both samples. Common engineered features were
dummy transformations for categorical features. A full-time/part-time flag, and a Rio Pace flag
were created. Variables derived from previous education and age. The latter was binned in
categorical ranges, while the former was used to create bins for different education type groups.
There were no missing values in the data, but some of the outliers for the continuous variables
were replaced using the first and third quartile whisker values for credits attempted, cumulative
earned hours, and cumulative quality hours. Frequencies and bar charts were undertaken to get a
better understanding of the data. Those results are not presented in this paper. Distributional plots
were also pursued for the continuous attributes. However, the research only presents bar, box,
plots, and densities relevant for the modelling process and final selection.
A visualization of the correlation matrix among the features selected for modelling is provided
below in figure 1. The correlation chart shows most variables are not strongly correlated among
themselves or to the outcome of interest, except for the fulltime flag and credits attempted at .69.
The variable cumulative quality hours is also correlated with cumulative earned hours at .57.
Negative correlation is present between the Rio PACE indicator and age younger than 22 years,
at .52. The rest of the coefficients are below .5.
7. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
7
Figure1. Correlation Matrix for Selected Features
Boxplots are also provided for the continuous variables -credits attempted, credits earned, and
term GPA without outliers. There appears to be statistical differences among the variables
between persisting and non-persisting students in those three variables. In all instances, persisting
students take and earn more credits. Subsequently, they manage to achieve a higher term GPA.
The median values for credits attempted and earned and term GPA are larger for persisting than
for non-persisting students. Because the continuous variables appear to be non-normally
distributed, statistical testing was conducted using the Wilcoxon non-parametric test for those
variables at alpha=.05. The tests concluded that there were statistical differences in the means of
persisting and non-persisting students. Not shown in the analysis were the Chi-square testing for
the categorical variables with respect to the persistence indicator. All variables were statistically
significant suggesting that their inclusion in the modelling effort was appropriate. Also, collinear
diagnostics were conducted using the variance proportion factors and the condition index using
SAS.
8. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
8
Figure 2. Boxplots for Credits and Term GPA
These results are relevant to instructors and advisers because they signal student at risk of not
persisting. It is worth mentioning that as part of the exploratory analysis numerical variables
included in the model were treated for outliers and credits attempted was logarithmically
transformed.
Figure 3. Credits and Term GPA
Density plots are shown in figure 3 above, the credits attempted chart suggests that non-persisting
attempt about 3 credits more often than persisting students, but after about 6 credits non-
persisting students lag persisting ones. The second chart shows that persisting students earn more
credits than non-persisting students also. The skewed tail exists because some students are
enrolled in law enforcement training and take more courses than regular degree and non-degree
seeking students. For term GPA, the left tail of the distribution shows that non-persisting students
have problems maintaining a passing GPA, while the right tail demonstrated that persisting
students outperform non-persisting ones at around 2.5 GPA or higher. Thus, the chart strongly
suggests that term GPA could influence the student decision on whether to continue his/her
studies into the next term. This fact is likely to be reflected in the model result and in some of the
literature reviewed.
9. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
9
These descriptive findings will help strengthen the case for building a machine learning model
for persistence. The following section presents model results, all models estimated are visually
compared on key performance metrics: confusion metrics, feature importance, AUC, accuracy,
precision, recall, and F1-score. Based on the results, emphasis is given to results for the gradient
boosting machine (GBM) model.
5. EMPIRICAL RESULTS
To analyse persistence into the next term at Rio Salado college, we undertook various alternative
models: logistic regression, decision trees, bagging, ADABOOST, random forest, gradient
boosting machine, and XGBOOST. Ten-fold cross-validation comparison results suggested that
decision trees and bagging were the worst performers. The rest of the models are compared in
their performance metrics such as feature importance, area under the curve, accuracy, precision,
recall, and F1-score. A final model will be selected based on these metrics for possible
implementation and to provide prediction probability scores for current students.
Table 1. Models Comparison Results
(a) Training of Models
Metric
Logistic
Regression
Gradient
Boost
Tuned
XGBOOST
Tuned
ADABOOST
Tuned
Random
Forest
Tuned
Accuracy 0.7369 0.8627 0.6723 0.6980 0.8557
Recall 0.7891 0.8753 0.9944 0.8726 0.8659
Precision 0.7354 0.8676 0.6196 0.6646 0.8631
F1 0.7613 0.8714 0.7635 0.7545 0.8645
(b) Validation of Models
Metric
Logistic
Regression
Gradient
Boost
Tuned
XGBOOST
Tuned
ADABOOST
Tuned
Random
Forest
Tuned
Accuracy 0.7433 0.7903 0.6477 0.7002 0.7824
Recall 0.7860 0.8029 0.9787 0.8739 0.7932
Precision 0.7453 0.8029 0.6042 0.6663 0.7967
F1 0.7651 0.8029 0.7472 0.7561 0.7950
Tables (a) and (b) provide the set of performance metrics to be compared for each alternative
model for training and validation. Sample sizes were: 15567 and 3892, respectively. Often for
decision tree-based models, estimates for the training data set tend to overfit the data. In these
results only, the random forest model appears to suffer from slightly overfitting. For this reason,
training results are not very useful in making a model selection. It is necessary to evaluate
validation results as they tend to simulate environment production conditions. Training results in
table (b) suggest that both random forest tuned and gradient boost tuned have similar
performance with a slight advantage for the gradient boost tuned model.
10. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
10
The gradient boost tuned managed to achieve better performance metrics than all the models.
Recall, precision, and F1-score are about 80%, while accuracy is at 79%.
Table (a) and figure (4) demonstrate that random forest and gradient boosting tuned have similar
performance metrics; behind these models are the results for the logistic regression classifier.
Interestingly, XGBOOST achieved the highest recall but its performance decays in the rest of the
metrics. The gradient boosting model metrics remain in the range of 80%, except for accuracy at
about 79%. The random forest follows with slightly lower metrics in the range of 79% with a
similar value for recall. Logistic regression performance is slightly lower in the 74- 78% range.
Its recall value is at 78%. XGBOOST and ADABOOST reached the highest recall rate, 98 and
87%, respectively, but their performance in other metrics were lower.
Figure 4. Model Metrics Performance on Validation Data
For those reasons, the researcher is inclined to suggest three models to test for possible
implementation: gradient booster, random forest, and logistic regression. Next, we present the
AUC curves for these tree models to show why the gradient boot machine is the preferred model
of choice. Validation results for the three models are overlaid in the same chart. The farther to the
left the better the model classifies the validation data. The 45-degree line represents business as
usual, a 50-50 chance that someone persists or not.
Figure 5. ROC Curve Validation Results Comparison
11. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
11
In figure5 above two models have very close AUC performance. Gradient boost machine and the
random forest in their tuned version achieved 86 and 85% values, accordingly. The logistic
model while also having a decent performance coefficient, it trails both models with an AUC
value of 80%. Ideally, the closer the curves are to the upper left corner, the better the model is in
its classification of the outcome.
Next, we present both the confusion matrix, figure5 below, and the feature importance chart,
figure6, for the gradient boosting machine model. The true label represents the actual cases in the
validation dataset, while the predicted values are the predicted assignment of those cases. The
latter, gives a good approximation of model performance in a production-like environment. By
assessing the misclassification values in the confusion matrix, one is led to conclude that this
model does discriminate better between persisting and non-persisting students.
The main diagonal figure demonstrates the correct classification for non-persisting and persisting
students. One can observe the model does have a higher classification of persisting students
compared to non-persisting ones, i.e., 42.7% versus 36.3%.
Regarding the off-diagonal values instructors and advisors could be interested in the false
positive rates as this figure would represent cases that are not persisting identified as persisting.
Doing so, could lead to non-persistence prevention if accompanied by an intervention targeting
student at risk of not persisting. On the other hand, false negative rates create a false alarm and
could cause misallocation of scarce resources since these students are likely to persist with
minimal or no intervention.
There is no harm in the misclassification of persisting students as non-persisting, however there
will be resource misallocation if targeted for interventions such as intrusive advising or additional
student support. On the contrary, identifying non-persisting student as persisting could be a
missing opportunity, especially at a time when 2-year colleges are experiencing enrolment
declines.
Figure6. GBM, Confusion Matrix
Finally, we think it is important to illustrate what factors contribute to persistence. To address this
issue, the feature importance chart is presented below.
12. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
12
Figure 7. GBM Feature Importance
Ranked by importance one can observe that term GPA is the most important attribute that
influences whether a student will enrol into the next term, its impact is over 40%. Next, log
transformed credits attempted is the second most important feature with a coefficient on around
14%. Quality cumulative hours, cumulative earned hours, and age indicator of 22 years or less
have around 10% contribution to the probability of the outcome of interest. After that the rest of
the features have importance coefficients at 5% or lower.
Overall, one can observe that the variables belong to three groups: academic performance,
education background, and age. In the next section, we extend the support for possible model
implementation to the gradient boosting machine model because it presents slightly better results
on misclassification rates.
6. LIMITATIONS OF THE STUDY
The study focused on estimating student persistence into the next term. It explored alternative
machine learning models and compared their key metrics. The data was split into training and
validation. However, having a third data set for testing would have made the results more robust
as it would have eliminated the possibility for data leakage [10]. The latter is a potential problem
arising from trying to improve the model to enhance validation results. Moreover, the number of
variables used in the modelling effort could be expanded by engineering additional features as
suggested by [4] and [28], that is something to be explored in the future. Efforts will be made to
implement and track model results in a production environment. Despite these limitations, the
model selected has higher metric values than some of the models cited in the literature review,
[2], [25].
There are other studies comparing machine learning models with different attributes selected.
[26], attempted to identify factors responsible for hearth failure. The outcome of interest indicator
had three levels: mild, moderate, and severe. The algorithms utilized were: CART, NN, and
SVM. In the end, the CART model outperformed the other models with an accuracy of 84%. The
comparison is similar but the models tested are not the same. Furthermore, this research is among
the few ones using machine learning models applied to community college data using readily
available open-source software such as Python and Jupyter notebooks as IDE.
13. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
13
7. CONCLUSIONS AND RECOMMENDATIONS
A data set was extracted for students who took courses in the Fall of 2021 and Spring 2022. The
purpose was to assess persistence into the next term. Seven classifiers were initially explored:
logistic regression, Bagging, Random Forest, GBM, ADABOOST, XGBOOST and a decision
tree. In the end, only five models appeared to be worth of additional exploration: logistic
regression, Gradient Boosting, random forest, ADABOOST and XGBOOST.
Model performance was compared across the five models on key performance metrics for the
training and validation datasets: accuracy, precision, recall, and F1-Score. In the training dataset,
the random forest classifier appeared to achieve higher metrics. The gradient boosting classifier
was second with lower metrics performance, while the logistic regression classifier performance
was third. For the validation dataset, the gradient boosting model secured higher values in all the
key metrics. This was followed by the forest random forest and logistic regression classifiers.
Therefore, results suggest that the gradient boosting model would be the model of choice for
possible implementation.
Of the models evaluated we selected two for consideration: the gradient boost (GBM) and the
logistic regression. The former performs the best of all models, but the latter is easier to interpret
and implement. In both instances term GPA appears as the most important feature in explaining
student persistence. Transformed credits attempted is the second most important variable in the
GBM model, while in the logit has the college not indicated as the second most relevant. The
third most relevant is the transformed cumulative quality hours in the GBM, while in the logit it
is no previous college experience. Confusion matrix results, performance metrics, AUC chart,
and variable importance led the researcher to recommend Gradient Boosting Machine as the
model choice.
REFERENCES
[1] Aceujo, Esteban M., Frech J., Ugalde, Araya, M. P., & Zafar B. (2020). The impact of COVID-19
on student experience and expectations: Evidence from a survey. Journal of Public Economics 191.
[2] Agnihotri, L. & A. Ott. (2014). Building A Student At-Risk Model: An End-to-End Perspective.
Proceedings of the 7th
International Conference on Educational Data Mining.
[3] Altig, Dave, Baker S., Barrero, J. M., Bloom, N., Bunn, P., Chen S., Davis, S. J., Leather, J., Meyer,
B., Mihaylov, E., Mizen, P., Parker, N., Renault, T., Smietanka, P., & Thwaits, G. (2020).
Economic Uncertainty Before and During the COID-19 Pandemic. Journal of Public Economics
191.
[4] Bird, A. Kelli, Castleman, L. B., Mabel, Z., & Song, Y. (2021). Bringing Transparency to Predictive
Analytics: A Systematic Comparison of Predictive Modeling Methods in Higher Education.
(EdWorking Paper: 21-438). nages
[5] Community college enrollment crisis? Historical Trends in Community College Enrollment.
AACC, 2019.
[6] Ekowo, M., Palmer, I. (2016). The Promises and Peril of Predictive Analytics in Higher Education.
A Landscape Analysis. New America, Oct. 2016 Fain, P. Top of the Mountain?
https://www.insidehighered.com/news/2011/12/21/community-college-enrollment-growth-ends.
[7] Feldman, M. J. (1993) Factors Associated with One-Year Retention in a Community College.
Research in Higher Education, Vol. 34, No 4, 1993.
[8] Fike, D. S. & Fike, R. (2008). Predictors of First-Year Student Retention in the Community College.
Volume 36, Number 2. October 2008, 68-88.
[9] Juszkiewicz, J. (2020, July). Trends in Community College Enrollment and Completion Data, Issue
6. Washington, DC: American Association of Community Colleges.
[10] Kapoor, S. & Narayanan, A. (2022). Leakage and the Reproducibility Crisis in ML-based Science.
arXiv:2207.07048v1 [cs.LG] 14 Jul 2022
14. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
14
[11] Kyriakides, G. & Margaritis, K. G. Hands-on Ensemble Learning with Python. Packt.
www.packt.com.
[12] Kimbark, K., Peters, M. L., Richardson, T. (2016). Effectiveness of the Student Success Course on
Persistence, Retention, Academic Achievement, and Student Engagement. Community College
Journal of Research and Practice. http://dx.doi.org/10.1080/10668926.2016.1166352
[13] Klempin, S., Grant, M., Ramos, M. (2018). Practitioner Perspectives on the Use of Predictive
Analytics in Targeted Advising for College Students. CCRC Working Paper No. 103. May, 2018.
[14] Lane Terralever. The Pandemic’s Impact on Higher Education Marketing in 2020 and Beyond.
https//www.laneterralever.com/industries/higher-education-marketing-agency/higher-education-
marketing-white-paper-pandemic-impact.
[15] Lopez-Wagner, M. C., Carollo, T., Shindledecker. Predictors of Retention: Identification of
Students At-Risk and Implementation of Continued Intervention Strategies.
[16] Inside Higher Ed. December, 2020. Higher Ed Faces Steep Cuts with Recent Oil Bust.
https://www.insidehighered.com/news/2020/12/16/higher-ed-faces-steep-cuts-recent-oil-bust
[17] Ryan, M. G., Improving Retention and Academic Achievement for First-Time Students at a Two-
Year College. Community College Journal of Research and Practice, 37: 131-134, 2013.
[18] Miller, J. E. & Berger, J. B. A Modified Model of College Student Persistence: Exploring the
Relationship Between Astin’s Theory of Involvement and Tinto’s Theory of Student Departure.
Journal of College Student Development; Jul/Aug 1997; 38, 4; ProQuest Education Journals pg.
387.
[19] Miller, T. E. and Herreid, C. Analysis of Variables to Predict First-Year Persistence Using Logistic
Regression Analysis at the University of South Florida, 2008.
[20] Mishra, P. (2022) Practical Explainable AI Using Python. https://doi.org/10.1007/978-1-4842-7158-
2.
[21] Nakajima, M., A., Dembo, M. H., Mossler, R. Student Persistence in Community Colleges.
Community College Journal of Research and Practice, 36: 591-613, 2012.
[22] N. S. C. Stay Informed with the Latest Enrollment Information. November, 2020.
https://nscresearchcenter.org/stay-informed/
[23] N. S. C. The Role of Community Colleges in Postsecondary Success: Community Colleges
Outcomes Report. nscresearchcenter.org.
[24] Ornelas, F. and Ordonez, C. (2017). Predicting Student Success: A Naïve Bayesian Application to
Community College Data. Tech Know Learn 22:299-315. DOI 10.1007/s10758-017-9334-z.
[25] Parvez, R. & Chowdhury, N. H. K. (2020). Economics of Student Retention Behavior in Higher
Education. Paper presented at the Agricultural and Applied Economics Association Meetings, 2020.
[26] Prudvi, P. S., Sharifahmadian, E. Applying Machine Learning Techniques to Find Important
Attributes for Heart Failure Severity Assessment. International Journal of Computer Science
Engineering Applications (IJCSEA) Vol 7, No 5, October 2017.
[27] Smith, V.S., Lange A., & Huston, D. R. (2012). Predictive Modeling to Forecast Student Outcomes
and Effective Interventions in Online Community College Courses. Journal of Asynchronous
Learning Networks. Vol. 16, Issue 3.
[28] Taylor, C., Veeramachaneni, K., O’Reilly, U. Likely to Stop? Predicting Stopout in Massive Open
Online Courses. arXiv:1408.3382v1 [cs.CY] 14 Aug. 2014.
[29] The Institute for College Access & Success (2019). Don’t Stop Improving: Supporting Data-Driven
Continuous Improvement in College Student Outcomes, March 2019.
[30] Tinto, V. (1988). Stages of Student Departure: Reflections on the Longitudinal Character of Student
Living. Journal of Higher Education, Jul. – Aug., 1988, Vol. 59, No. 4 (Jul. – Aug., 1988), pp. 438-
455.
15. International Journal of Computer Science, Engineering and Applications (IJCSEA) Vol. 13, No. 01, February 2023
15
AUTHORS
FERMIN ORNELAS holds a Ph.D. in Agricultural Economics from Texas A&M University, College
Station TX, USA, and a Certificate in Data Science and Business Analytics from the McCombs School of
Business at the University of Texas in Austin TX, USA.
He is currently a Sr. Research Analyst at Rio Salado College. His professional research interests include:
machine learning modelling applied to student behaviour and the financial industry, i.e. student success,
persistence, and retention, credit risk, and customer loyalty.