In this paper the authors proposed different Multilayer Perceptron Models (MLP) of artificial neural
networks (ANN) suitable for visual merchandising in Global Distribution (GDO) applications involving
supermarket product facing. The models are related to the prediction of different attributes concerning
mainly shelf product allocation applying times series forecasting approach. The study highlights the range
validity of the sales prediction by analysing different products allocated on a testing shelf. The paper shows
the correct procedures able to analyse most guaranteed results, by describing how test and train datasets
can be processed. The prediction results are useful in order to design monthly a planogram by taking into
account the shelf allocations, the general sales trend, and the promotion activities. The preliminary
correlation analysis provided an innovative key reading of the predicted outputs. The testing has been
performed by Weka and RapidMiner tools able to predict by MLP ANN each attribute of the experimental
dataset. Finally it is formulated an innovative hybrid model which combines Weka prediction outputs as
input of the MLP ANN RapidMiner algorithm. This implementation allows to use an artificial testing
dataset useful when experimental datasets are composed by few data, thus accelerating the self-learning
process of the model. The proposed study is developed within a framework of an industry project.
By applying RapidMiner workflows has been processed a dataset originated from different data files, and containing information about the sales over three years of a large chain of retail stores. Subsequently, has been constructed a Deep Learning model performing a predictive algorithm suitable for sales forecasting. This model is based on artificial neural network –ANN- algorithm able to learn the model starting from sales historical data and by pre-processing the data. The best built model uses a multilayer neural network together with an “optimized operator” able to find automatically the best parameter setting of the implemented algorithm. In order to prove the best performing predictive model, other machine learning algorithms have been tested. The performance comparison has been performed between Support Vector Machine –SVM-, k-Nearest Neighbor k-NN-,Gradient Boosted Trees, Decision Trees, and Deep Learning algorithms. The comparison of the degree of correlation between real and predicted values, the average absolute error and the relative average error proved that ANN exhibited the best performance. The Gradient Boosted Trees approach represents an alternative approach having the second best performance. The case of study has been developed within the framework of an industry project oriented on the integration of high performance data mining models able to predict sales using–ERP- and customer relationship management –CRM- tools.
DATA MINING MODEL PERFORMANCE OF SALES PREDICTIVE ALGORITHMS BASED ON RAPIDMI...ijcsit
By applying RapidMiner workflows has been processed a dataset originated from different data files, and containing information about the sales over three years of a large chain of retail stores. Subsequently, has been constructed a Deep Learning model performing a predictive algorithm suitable for sales forecasting. This model is based on artificial neural network –ANN- algorithm able to learn the model starting from
sales historical data and by pre-processing the data. The best built model uses a multilayer neural network together with an “optimized operator” able to find automatically the best parameter setting of the implemented algorithm. In order to prove the best performing predictive model, other machine learning algorithms have been tested. The performance comparison has been performed between Support Vector
Machine –SVM-, k-Nearest Neighbor k-NN-,Gradient Boosted Trees, Decision Trees, and Deep Learning algorithms. The comparison of the degree of correlation between real and predicted values, the average
absolute error and the relative average error proved that ANN exhibited the best performance. The Gradient Boosted Trees approach represents an alternative approach having the second best performance. The case of study has been developed within the framework of an industry project oriented on the
integration of high performance data mining models able to predict sales using–ERP- and customer relationship management –CRM- tools.
Visual and analytical mining of sales transaction data for production plannin...Gurdal Ertek
Recent developments in information technology paved the way for the collection of large amounts of data pertaining to various aspects of an enterprise. The greatest challenge faced in
processing these massive amounts of raw data gathered turns out to be the effective management of data with the ultimate purpose of deriving necessary and meaningful information
out of it. The following paper presents an attempt to illustrate the combination of visual and analytical data mining techniques for planning of marketing and production activities. The
primary phases of the proposed framework consist of filtering, clustering and comparison steps implemented using interactive pie charts, K-Means algorithm and parallel coordinate plots
respectively. A prototype decision support system is developed and a sample analysis session is conducted to demonstrate the applicability of the framework.
http://research.sabanciuniv.edu.
DATA MINING APPLIED IN FOOD TRADE NETWORKgerogepatton
The proposed study deals with the design and the development of a Decision Support System (DSS) platform suitable for the global distribution system (GDS). Precisely, the prototype platform combines artificial intelligence and data mining algorithms to process data collected into a Cassandra Big Data system. In the first part of the paper platform architectures together with all the adopted frameworks including Key Performance Indicators (KPIs) definitions and risk mapping design have been discussed. In the second part data mining algorithms have been applied in order to predict main KPIs. The adopted artificial neural networks architectures are Long Short-Term Memory (LSTM), standard Recurrent Neural Network (RNN) and Gated Recurrent Units (GRU). A dataset with KPIs has been generated in order to test the algorithms. All performed algorithms show a good matching with the generated dataset, thus proving to be the correct approach to predict KPIs. The best performances in terms of Accuracy and Loss are reached by using the standard RNN. The proposed platform represents a solution to increase the Knowledge Base (KB) for a strategic marketing and advanced business intelligence operations
The proposed study deals with the design and the development of a Decision Support System (DSS) platform suitable for the global distribution system (GDS). Precisely, the prototype platform combines artificial intelligence and data mining algorithms to process data collected into a Cassandra Big Data system. In the first part of the paper platform architectures together with all the adopted frameworks including Key Performance Indicators (KPIs) definitions and risk mapping design have been discussed. In the second part data mining algorithms have been applied in order to predict main KPIs. The adopted artificial neural networks architectures are Long Short-Term Memory (LSTM), standard Recurrent Neural Network (RNN) and Gated Recurrent Units (GRU). A dataset with KPIs has been generated in order to test the algorithms. All performed algorithms show a good matching with the generated dataset, thus proving to be the correct approach to predict KPIs. The best performances in terms of Accuracy and Loss are reached by using the standard RNN. The proposed platform represents a solution to increase the Knowledge Base (KB) for a strategic marketing and advanced business intelligence operations.
In this paper, feature tracking based and histogram based traffic congestion detection systems are developed. Developed all system are designed to run as real time application. In this work, ORB (Oriented FAST and Rotated BRIEF) feature extraction method have been used to develop feature tracking based traffic congestion solution. ORB is a rotation invariant, fast and resistant to noise method and contains the power of FAST and BRIEF feature extraction methods. Also, two different approaches, which are standard deviation and weighed average, have been applied to find out the congestion information by using histogram of the image to develop histogram based traffic congestion solution. Both systems have been tested on different weather conditions such as cloudy, sunny and rainy to provide various illumination at both daytime and night. For all developed systems performance results are examined to show the advantages and drawbacks of these systems.
APPLICATION OF FACEBOOK'S PROPHET ALGORITHM FOR SUCCESSFUL SALES FORECASTING ...ijnlc
This paper presents a framework capable of accurately forecasting future sales in the retail industry and classifying the product portfolio according to the expected level of forecasting reliability. The proposed framework, that would be of great use for any company operating in the retail industry, is based on Facebook's Prophet algorithm and backtesting strategy. Real-world sales forecasting benchmark data obtained experimentally in a production environment in one of the biggest retail companies in Bosnia and Herzegovina is used to evaluate the framework and demonstrate its capabilities in a real-world use case scenario
UNDERSTANDING THE APPLICABILITY OF LINEAR & NON-LINEAR MODELS USING A CASE-BA...ijaia
This paper uses a case based study – “product sales estimation” on real-time data to help us understand
the applicability of linear and non-linear models in machine learning and data mining. A systematic
approach has been used here to address the given problem statement of sales estimation for a particular set
of products in multiple categories by applying both linear and non-linear machine learning techniques on
a data set of selected features from the original data set. Feature selection is a process that reduces the
dimensionality of the data set by excluding those features which contribute minimal to the prediction of the
dependent variable. The next step in this process is training the model that is done using multiple
techniques from linear & non-linear domains, one of the best ones in their respective areas. Data Remodeling
has then been done to extract new features from the data set by changing the structure of the
dataset & the performance of the models is checked again. Data Remodeling often plays a very crucial and
important role in boosting classifier accuracies by changing the properties of the given dataset. We then try
to explore and analyze the various reasons due to which one model performs better than the other & hence
try and develop an understanding about the applicability of linear & non-linear machine learning models.
The target mentioned above being our primary goal, we also aim to find the classifier with the best possible
accuracy for product sales estimation in the given scenario.
By applying RapidMiner workflows has been processed a dataset originated from different data files, and containing information about the sales over three years of a large chain of retail stores. Subsequently, has been constructed a Deep Learning model performing a predictive algorithm suitable for sales forecasting. This model is based on artificial neural network –ANN- algorithm able to learn the model starting from sales historical data and by pre-processing the data. The best built model uses a multilayer neural network together with an “optimized operator” able to find automatically the best parameter setting of the implemented algorithm. In order to prove the best performing predictive model, other machine learning algorithms have been tested. The performance comparison has been performed between Support Vector Machine –SVM-, k-Nearest Neighbor k-NN-,Gradient Boosted Trees, Decision Trees, and Deep Learning algorithms. The comparison of the degree of correlation between real and predicted values, the average absolute error and the relative average error proved that ANN exhibited the best performance. The Gradient Boosted Trees approach represents an alternative approach having the second best performance. The case of study has been developed within the framework of an industry project oriented on the integration of high performance data mining models able to predict sales using–ERP- and customer relationship management –CRM- tools.
DATA MINING MODEL PERFORMANCE OF SALES PREDICTIVE ALGORITHMS BASED ON RAPIDMI...ijcsit
By applying RapidMiner workflows has been processed a dataset originated from different data files, and containing information about the sales over three years of a large chain of retail stores. Subsequently, has been constructed a Deep Learning model performing a predictive algorithm suitable for sales forecasting. This model is based on artificial neural network –ANN- algorithm able to learn the model starting from
sales historical data and by pre-processing the data. The best built model uses a multilayer neural network together with an “optimized operator” able to find automatically the best parameter setting of the implemented algorithm. In order to prove the best performing predictive model, other machine learning algorithms have been tested. The performance comparison has been performed between Support Vector
Machine –SVM-, k-Nearest Neighbor k-NN-,Gradient Boosted Trees, Decision Trees, and Deep Learning algorithms. The comparison of the degree of correlation between real and predicted values, the average
absolute error and the relative average error proved that ANN exhibited the best performance. The Gradient Boosted Trees approach represents an alternative approach having the second best performance. The case of study has been developed within the framework of an industry project oriented on the
integration of high performance data mining models able to predict sales using–ERP- and customer relationship management –CRM- tools.
Visual and analytical mining of sales transaction data for production plannin...Gurdal Ertek
Recent developments in information technology paved the way for the collection of large amounts of data pertaining to various aspects of an enterprise. The greatest challenge faced in
processing these massive amounts of raw data gathered turns out to be the effective management of data with the ultimate purpose of deriving necessary and meaningful information
out of it. The following paper presents an attempt to illustrate the combination of visual and analytical data mining techniques for planning of marketing and production activities. The
primary phases of the proposed framework consist of filtering, clustering and comparison steps implemented using interactive pie charts, K-Means algorithm and parallel coordinate plots
respectively. A prototype decision support system is developed and a sample analysis session is conducted to demonstrate the applicability of the framework.
http://research.sabanciuniv.edu.
DATA MINING APPLIED IN FOOD TRADE NETWORKgerogepatton
The proposed study deals with the design and the development of a Decision Support System (DSS) platform suitable for the global distribution system (GDS). Precisely, the prototype platform combines artificial intelligence and data mining algorithms to process data collected into a Cassandra Big Data system. In the first part of the paper platform architectures together with all the adopted frameworks including Key Performance Indicators (KPIs) definitions and risk mapping design have been discussed. In the second part data mining algorithms have been applied in order to predict main KPIs. The adopted artificial neural networks architectures are Long Short-Term Memory (LSTM), standard Recurrent Neural Network (RNN) and Gated Recurrent Units (GRU). A dataset with KPIs has been generated in order to test the algorithms. All performed algorithms show a good matching with the generated dataset, thus proving to be the correct approach to predict KPIs. The best performances in terms of Accuracy and Loss are reached by using the standard RNN. The proposed platform represents a solution to increase the Knowledge Base (KB) for a strategic marketing and advanced business intelligence operations
The proposed study deals with the design and the development of a Decision Support System (DSS) platform suitable for the global distribution system (GDS). Precisely, the prototype platform combines artificial intelligence and data mining algorithms to process data collected into a Cassandra Big Data system. In the first part of the paper platform architectures together with all the adopted frameworks including Key Performance Indicators (KPIs) definitions and risk mapping design have been discussed. In the second part data mining algorithms have been applied in order to predict main KPIs. The adopted artificial neural networks architectures are Long Short-Term Memory (LSTM), standard Recurrent Neural Network (RNN) and Gated Recurrent Units (GRU). A dataset with KPIs has been generated in order to test the algorithms. All performed algorithms show a good matching with the generated dataset, thus proving to be the correct approach to predict KPIs. The best performances in terms of Accuracy and Loss are reached by using the standard RNN. The proposed platform represents a solution to increase the Knowledge Base (KB) for a strategic marketing and advanced business intelligence operations.
In this paper, feature tracking based and histogram based traffic congestion detection systems are developed. Developed all system are designed to run as real time application. In this work, ORB (Oriented FAST and Rotated BRIEF) feature extraction method have been used to develop feature tracking based traffic congestion solution. ORB is a rotation invariant, fast and resistant to noise method and contains the power of FAST and BRIEF feature extraction methods. Also, two different approaches, which are standard deviation and weighed average, have been applied to find out the congestion information by using histogram of the image to develop histogram based traffic congestion solution. Both systems have been tested on different weather conditions such as cloudy, sunny and rainy to provide various illumination at both daytime and night. For all developed systems performance results are examined to show the advantages and drawbacks of these systems.
APPLICATION OF FACEBOOK'S PROPHET ALGORITHM FOR SUCCESSFUL SALES FORECASTING ...ijnlc
This paper presents a framework capable of accurately forecasting future sales in the retail industry and classifying the product portfolio according to the expected level of forecasting reliability. The proposed framework, that would be of great use for any company operating in the retail industry, is based on Facebook's Prophet algorithm and backtesting strategy. Real-world sales forecasting benchmark data obtained experimentally in a production environment in one of the biggest retail companies in Bosnia and Herzegovina is used to evaluate the framework and demonstrate its capabilities in a real-world use case scenario
UNDERSTANDING THE APPLICABILITY OF LINEAR & NON-LINEAR MODELS USING A CASE-BA...ijaia
This paper uses a case based study – “product sales estimation” on real-time data to help us understand
the applicability of linear and non-linear models in machine learning and data mining. A systematic
approach has been used here to address the given problem statement of sales estimation for a particular set
of products in multiple categories by applying both linear and non-linear machine learning techniques on
a data set of selected features from the original data set. Feature selection is a process that reduces the
dimensionality of the data set by excluding those features which contribute minimal to the prediction of the
dependent variable. The next step in this process is training the model that is done using multiple
techniques from linear & non-linear domains, one of the best ones in their respective areas. Data Remodeling
has then been done to extract new features from the data set by changing the structure of the
dataset & the performance of the models is checked again. Data Remodeling often plays a very crucial and
important role in boosting classifier accuracies by changing the properties of the given dataset. We then try
to explore and analyze the various reasons due to which one model performs better than the other & hence
try and develop an understanding about the applicability of linear & non-linear machine learning models.
The target mentioned above being our primary goal, we also aim to find the classifier with the best possible
accuracy for product sales estimation in the given scenario.
AUTOMATION OF BEST-FIT MODEL SELECTION USING A BAG OF MACHINE LEARNING LIBRAR...ijaia
Sales forecasting became crucial for industries in past decades with rapid globalization, widespread adoption of information technology towards e-business, understanding market fluctuations, meeting business plans, and avoiding loss of sales. This research precisely predicts the automotive industry sales using a bag of multiple machine learning and time series algorithms coupled with historical sales and auxiliary features. Three-year historical sales data (from 2017 till 2020) were used for the model building or training, and one-year (2020-2021) predictions were computed for 900 unique SKU's (stock-keeping units). In the present study, the SKU is a combination of sales office, core business field, and material customer group. Various data cleaning and exploratory data analysis algorithms were implemented over raw datasets before use for modeling. Mean absolute percentage error (mape) were estimated for individual predictions from time series and machine learning models. The best model was selected for unique SKU's as per the most negligible mape value.
INNOVATIVE BI APPROACHES AND METHODOLOGIES IMPLEMENTING A MULTILEVEL ANALYTIC...ijscai
The paper proposes an advanced Multilevel Analytics Model –MAM-, applied on a specific case of study
referring to a research project involving an industry mainly working in roadside assistance service (ACI
Global S.p.A.). In the first part of the paper are described the initial specifications of the research project by
addressing the study on information system architectures explaining knowledge gain, decision making and
data flow automatism applied on the specific case of study. In the second part of the paper is described in
details the MAM acting on different analytics levels, by describing the first analyzer module and the second
one involving data mining and analytical model suitable for strategic marketing e business intelligence –BI-.
The analyzer module is represented by graphical dashboards useful to understand the industry business trend
and to execute main decision making. The second module is suitable to understand deeply services trend and
clustering by finding possible correlations between the variables to analyze. For the data mining processing
has been applied the Rapid Miner workflows of K-Means clustering and the Correlation Matrix. The second
part of the paper is mainly focused on analytical models representing phenomena such as vehicle accidents
and fleet car sharing trend, which can be correlated with strategic car services. The proposed architectures
and models represent methodologies and approaches able to improve strategic marketing and –BI- advanced
analytics following ‘Frascati’ research guidelines. The MAM model can be adopted for other cases of study
concerning other industry applications. The originality of the paper is the scientific methodological approach
used to interpret and read data, by executing advanced analytical models based on data mining algorithms
which can be applied on industry database systems representing knowledge base.
The usage of Neural network s has determined a variegated area of packages in the present world. This has caused the
improvement of various fashions for economic markets and funding. This paper represents the idea the way to predict share
market fee the use of artificial Neural community with a given enter parameters of share marketplace. The proportion
marketplace is dynamic in nature approach to expect percentage fee could be very complex method by using trendy prediction
or computation method. Its predominant motive is that there is no linear relationship between market parameters and target last
price. Since there is no linear relationship between input patterns and corresponding output patterns, so use of neural network is
a desire of hobby for share market prediction.
Framework to Analyze Customer’s Feedback in Smartphone Industry Using Opinion...IJECEIAES
In the present age, cellular phones are the largest selling products in the world. Big Data Analytics is a method used for examining large and varied data, which we know as big data. Big data analytics is very useful for understanding the world of cellphone business. It is important to understand the requirements, demands, and opinions of the customer. Opinion Mining is getting more important than ever before, for performing analysis and forecasting customer behavior and preferences. This study proposes a framework about the key features of cellphones based on which, customers buy them and rate them accordingly. This research work also provides balanced and well researched reasons as to why few companies enjoy dominance in the market, while others do not make as much of an impact.
Analytics are forming the basis of competition today. This white paper addresses what distinguishes analytics and answers the question are we doing analytics in the data warehouse? The paper further talks about the contending Platforms for the Analytics Workload and introduces the ParAccel Analytic Database as a key component of Information Architecture.
A CASE STUDY OF INNOVATION OF AN INFORMATION COMMUNICATION SYSTEM AND UPGRADE...ijaia
In this paper, a case study is analyzed. This case study is about an upgrade of an industry communication system developed by following Frascati research guidelines. The knowledge Base (KB) of the industry is gained by means of different tools that are able to provide data and information having different formats and structures into an unique bus system connected to a Big Data. The initial part of the research is focused on the implementation of strategic tools, which can able to upgrade the KB. The second part of the proposed study is related to the implementation of innovative algorithms based on a KNIME (Konstanz Information Miner) Gradient Boosted Trees workflow processing data of the communication system which travel into an Enterprise Service Bus (ESB) infrastructure. The goal of the paper is to prove that all the new KB collected into a Cassandra big data system could be processed through the ESB by predictive algorithms solving possible conflicts between hardware and software. The conflicts are due to the integration of different database technologies and data structures. In order to check the outputs of the Gradient Boosted Trees algorithm an experimental dataset suitable for machine learning testing has been tested. The test has been performed on a prototype network system modeling a part of the whole communication system. The paper shows how to validate industrial research by following a complete design and development of a whole communication system network improving business intelligence (BI).
An Empirical Study of the Improved SPLD Framework using Expert Opinion TechniqueIJEACS
Due to the growing need for high-performance and low-cost software applications and the increasing competitiveness, the industry is under pressure to deliver products with low development cost, reduced delivery time and improved quality. To address these demands, researchers have proposed several development methodologies and frameworks. One of the latest methodologies is software product line (SPL) which utilizes the concepts like reusability and variability to deliver successful products with shorter time-to-market, least development and minimum maintenance cost with a high-quality product. This research paper is a validation of our proposed framework, Improved Software Product Line (ISPL), using Expert Opinion Technique. An extensive survey based on a set of questionnaires on various aspects and sub-processes of the ISPLD Framework was carried. Analysis of the empirical data concludes that ISPL shows significant improvements on several aspects of the contemporary SPL frameworks.
Survey of the Euro Currency Fluctuation by Using Data Miningijcsit
Data mining or Knowledge Discovery in Databases (KDD) is a new field in information technology that emerged because of progress in creation and maintenance of large databases by combining statistical and artificial intelligence methods with database management. Data mining is used to recognize hidden patterns and provide relevant information for decision making on complex problems where conventional methods are inecient or too slow. Data mining can be used as a powerful tool to predict future trends and behaviors, and this prediction allows making proactive, knowledge-driven decisions in businesses. Since the automated prospective analyses offered by data mining move beyond the analyses of past events provided by retrospective tools, it can answer the business questions which are traditionally time consuming to resolve. Based on this great advantage, it provides more interest for the government, industry and commerce. In this paper we have used this tool to investigate the Euro currency fluctuation.For this investigation, we have three different algorithms: K*, IBK and MLP and we have extracted.Euro currency volatility by using the same criteria for all used algorithms. The used dataset has
21,084 records and is collected from daily price fluctuations in the Euro currency in the period
of10/2006 to 04/2010.
McKinsey Global Institute Big data The next frontier for innova.docxandreecapon
McKinsey Global Institute
Big data: The next frontier for innovation, competition, and productivity 27
2. Bigdatatechniquesand technologies
A wide variety of techniques and technologies has been developed and adapted to aggregate, manipulate, analyze, and visualize big data. These techniques and technologies draw from several fields including statistics, computer science, applied mathematics, and economics. This means that an organization that intends to derive value from big data has to adopt a flexible, multidisciplinary approach. Some techniques and technologies were developed in a world with access to far smaller volumes and variety in data, but have been successfully adapted so that they are applicable to very large sets of more diverse data. Others have been developed more recently, specifically to capture value from big data. Some were developed by academics and others by companies, especially those with online business models predicated on analyzing big data.
This report concentrates on documenting the potential value that leveraging big data can create. It is not a detailed instruction manual on how to capture value, a task that requires highly specific customization to an organization’s context, strategy, and capabilities. However, we wanted to note some of the main techniques and technologies that can be applied to harness big data to clarify the way some
of the levers for the use of big data that we describe might work. These are not comprehensive lists—the story of big data is still being written; new methods and tools continue to be developed to solve new problems. To help interested readers find a particular technique or technology easily, we have arranged these lists alphabetically. Where we have used bold typefaces, we are illustrating the multiple interconnections between techniques and technologies. We also provide a brief selection of illustrative examples of visualization, a key tool for understanding very large-scale data and complex analyses in order to make better decisions.
TECHNIQUES FOR ANALYZING BIG DATA
There are many techniques that draw on disciplines such as statistics and computer science (particularly machine learning) that can be used to analyze datasets. In this section, we provide a list of some categories of techniques applicable across a range of industries. This list is by no means exhaustive. Indeed, researchers continue to develop new techniques and improve on existing ones, particularly in response to the need
to analyze new combinations of data. We note that not all of these techniques strictly require the use of big data—some of them can be applied effectively to smaller datasets (e.g., A/B testing, regression analysis). However, all of the techniques we list here can be applied to big data and, in general, larger and more diverse datasets can be used to generate more numerous and insightful results than smaller, less diverse ones.
A/B testing. A technique in which a control group is compa ...
Visual and analytical mining of transactions data for production planning f...ertekg
Download Link > https://ertekprojects.com/gurdal-ertek-publications/blog/visual-and-analytical-mining-of-sales-transaction-data-for-production-planning-and-marketing/
Recent developments in information technology paved the way for the collection of large amounts of data pertaining to various aspects of an enterprise. The greatest challenge faced in processing these massive amounts of raw data gathered turns out to be the effective management of data with the ultimate purpose of deriving necessary and meaningful information out of it. The following paper presents an attempt to illustrate the combination of visual and analytical data mining techniques for planning of marketing and production activities. The primary phases of the proposed framework consist of filtering, clustering and comparison steps
implemented using interactive pie charts, K-Means algorithm and parallel coordinate plots respectively. A prototype decision support system is developed and a sample analysis session is conducted to demonstrate the applicability of the framework.
AUTOMATION OF BEST-FIT MODEL SELECTION USING A BAG OF MACHINE LEARNING LIBRAR...ijaia
Sales forecasting became crucial for industries in past decades with rapid globalization, widespread adoption of information technology towards e-business, understanding market fluctuations, meeting business plans, and avoiding loss of sales. This research precisely predicts the automotive industry sales using a bag of multiple machine learning and time series algorithms coupled with historical sales and auxiliary features. Three-year historical sales data (from 2017 till 2020) were used for the model building or training, and one-year (2020-2021) predictions were computed for 900 unique SKU's (stock-keeping units). In the present study, the SKU is a combination of sales office, core business field, and material customer group. Various data cleaning and exploratory data analysis algorithms were implemented over raw datasets before use for modeling. Mean absolute percentage error (mape) were estimated for individual predictions from time series and machine learning models. The best model was selected for unique SKU's as per the most negligible mape value.
INNOVATIVE BI APPROACHES AND METHODOLOGIES IMPLEMENTING A MULTILEVEL ANALYTIC...ijscai
The paper proposes an advanced Multilevel Analytics Model –MAM-, applied on a specific case of study
referring to a research project involving an industry mainly working in roadside assistance service (ACI
Global S.p.A.). In the first part of the paper are described the initial specifications of the research project by
addressing the study on information system architectures explaining knowledge gain, decision making and
data flow automatism applied on the specific case of study. In the second part of the paper is described in
details the MAM acting on different analytics levels, by describing the first analyzer module and the second
one involving data mining and analytical model suitable for strategic marketing e business intelligence –BI-.
The analyzer module is represented by graphical dashboards useful to understand the industry business trend
and to execute main decision making. The second module is suitable to understand deeply services trend and
clustering by finding possible correlations between the variables to analyze. For the data mining processing
has been applied the Rapid Miner workflows of K-Means clustering and the Correlation Matrix. The second
part of the paper is mainly focused on analytical models representing phenomena such as vehicle accidents
and fleet car sharing trend, which can be correlated with strategic car services. The proposed architectures
and models represent methodologies and approaches able to improve strategic marketing and –BI- advanced
analytics following ‘Frascati’ research guidelines. The MAM model can be adopted for other cases of study
concerning other industry applications. The originality of the paper is the scientific methodological approach
used to interpret and read data, by executing advanced analytical models based on data mining algorithms
which can be applied on industry database systems representing knowledge base.
The usage of Neural network s has determined a variegated area of packages in the present world. This has caused the
improvement of various fashions for economic markets and funding. This paper represents the idea the way to predict share
market fee the use of artificial Neural community with a given enter parameters of share marketplace. The proportion
marketplace is dynamic in nature approach to expect percentage fee could be very complex method by using trendy prediction
or computation method. Its predominant motive is that there is no linear relationship between market parameters and target last
price. Since there is no linear relationship between input patterns and corresponding output patterns, so use of neural network is
a desire of hobby for share market prediction.
Framework to Analyze Customer’s Feedback in Smartphone Industry Using Opinion...IJECEIAES
In the present age, cellular phones are the largest selling products in the world. Big Data Analytics is a method used for examining large and varied data, which we know as big data. Big data analytics is very useful for understanding the world of cellphone business. It is important to understand the requirements, demands, and opinions of the customer. Opinion Mining is getting more important than ever before, for performing analysis and forecasting customer behavior and preferences. This study proposes a framework about the key features of cellphones based on which, customers buy them and rate them accordingly. This research work also provides balanced and well researched reasons as to why few companies enjoy dominance in the market, while others do not make as much of an impact.
Analytics are forming the basis of competition today. This white paper addresses what distinguishes analytics and answers the question are we doing analytics in the data warehouse? The paper further talks about the contending Platforms for the Analytics Workload and introduces the ParAccel Analytic Database as a key component of Information Architecture.
A CASE STUDY OF INNOVATION OF AN INFORMATION COMMUNICATION SYSTEM AND UPGRADE...ijaia
In this paper, a case study is analyzed. This case study is about an upgrade of an industry communication system developed by following Frascati research guidelines. The knowledge Base (KB) of the industry is gained by means of different tools that are able to provide data and information having different formats and structures into an unique bus system connected to a Big Data. The initial part of the research is focused on the implementation of strategic tools, which can able to upgrade the KB. The second part of the proposed study is related to the implementation of innovative algorithms based on a KNIME (Konstanz Information Miner) Gradient Boosted Trees workflow processing data of the communication system which travel into an Enterprise Service Bus (ESB) infrastructure. The goal of the paper is to prove that all the new KB collected into a Cassandra big data system could be processed through the ESB by predictive algorithms solving possible conflicts between hardware and software. The conflicts are due to the integration of different database technologies and data structures. In order to check the outputs of the Gradient Boosted Trees algorithm an experimental dataset suitable for machine learning testing has been tested. The test has been performed on a prototype network system modeling a part of the whole communication system. The paper shows how to validate industrial research by following a complete design and development of a whole communication system network improving business intelligence (BI).
An Empirical Study of the Improved SPLD Framework using Expert Opinion TechniqueIJEACS
Due to the growing need for high-performance and low-cost software applications and the increasing competitiveness, the industry is under pressure to deliver products with low development cost, reduced delivery time and improved quality. To address these demands, researchers have proposed several development methodologies and frameworks. One of the latest methodologies is software product line (SPL) which utilizes the concepts like reusability and variability to deliver successful products with shorter time-to-market, least development and minimum maintenance cost with a high-quality product. This research paper is a validation of our proposed framework, Improved Software Product Line (ISPL), using Expert Opinion Technique. An extensive survey based on a set of questionnaires on various aspects and sub-processes of the ISPLD Framework was carried. Analysis of the empirical data concludes that ISPL shows significant improvements on several aspects of the contemporary SPL frameworks.
Survey of the Euro Currency Fluctuation by Using Data Miningijcsit
Data mining or Knowledge Discovery in Databases (KDD) is a new field in information technology that emerged because of progress in creation and maintenance of large databases by combining statistical and artificial intelligence methods with database management. Data mining is used to recognize hidden patterns and provide relevant information for decision making on complex problems where conventional methods are inecient or too slow. Data mining can be used as a powerful tool to predict future trends and behaviors, and this prediction allows making proactive, knowledge-driven decisions in businesses. Since the automated prospective analyses offered by data mining move beyond the analyses of past events provided by retrospective tools, it can answer the business questions which are traditionally time consuming to resolve. Based on this great advantage, it provides more interest for the government, industry and commerce. In this paper we have used this tool to investigate the Euro currency fluctuation.For this investigation, we have three different algorithms: K*, IBK and MLP and we have extracted.Euro currency volatility by using the same criteria for all used algorithms. The used dataset has
21,084 records and is collected from daily price fluctuations in the Euro currency in the period
of10/2006 to 04/2010.
McKinsey Global Institute Big data The next frontier for innova.docxandreecapon
McKinsey Global Institute
Big data: The next frontier for innovation, competition, and productivity 27
2. Bigdatatechniquesand technologies
A wide variety of techniques and technologies has been developed and adapted to aggregate, manipulate, analyze, and visualize big data. These techniques and technologies draw from several fields including statistics, computer science, applied mathematics, and economics. This means that an organization that intends to derive value from big data has to adopt a flexible, multidisciplinary approach. Some techniques and technologies were developed in a world with access to far smaller volumes and variety in data, but have been successfully adapted so that they are applicable to very large sets of more diverse data. Others have been developed more recently, specifically to capture value from big data. Some were developed by academics and others by companies, especially those with online business models predicated on analyzing big data.
This report concentrates on documenting the potential value that leveraging big data can create. It is not a detailed instruction manual on how to capture value, a task that requires highly specific customization to an organization’s context, strategy, and capabilities. However, we wanted to note some of the main techniques and technologies that can be applied to harness big data to clarify the way some
of the levers for the use of big data that we describe might work. These are not comprehensive lists—the story of big data is still being written; new methods and tools continue to be developed to solve new problems. To help interested readers find a particular technique or technology easily, we have arranged these lists alphabetically. Where we have used bold typefaces, we are illustrating the multiple interconnections between techniques and technologies. We also provide a brief selection of illustrative examples of visualization, a key tool for understanding very large-scale data and complex analyses in order to make better decisions.
TECHNIQUES FOR ANALYZING BIG DATA
There are many techniques that draw on disciplines such as statistics and computer science (particularly machine learning) that can be used to analyze datasets. In this section, we provide a list of some categories of techniques applicable across a range of industries. This list is by no means exhaustive. Indeed, researchers continue to develop new techniques and improve on existing ones, particularly in response to the need
to analyze new combinations of data. We note that not all of these techniques strictly require the use of big data—some of them can be applied effectively to smaller datasets (e.g., A/B testing, regression analysis). However, all of the techniques we list here can be applied to big data and, in general, larger and more diverse datasets can be used to generate more numerous and insightful results than smaller, less diverse ones.
A/B testing. A technique in which a control group is compa ...
Visual and analytical mining of transactions data for production planning f...ertekg
Download Link > https://ertekprojects.com/gurdal-ertek-publications/blog/visual-and-analytical-mining-of-sales-transaction-data-for-production-planning-and-marketing/
Recent developments in information technology paved the way for the collection of large amounts of data pertaining to various aspects of an enterprise. The greatest challenge faced in processing these massive amounts of raw data gathered turns out to be the effective management of data with the ultimate purpose of deriving necessary and meaningful information out of it. The following paper presents an attempt to illustrate the combination of visual and analytical data mining techniques for planning of marketing and production activities. The primary phases of the proposed framework consist of filtering, clustering and comparison steps
implemented using interactive pie charts, K-Means algorithm and parallel coordinate plots respectively. A prototype decision support system is developed and a sample analysis session is conducted to demonstrate the applicability of the framework.
Final project report on grocery store management system..pdfKamal Acharya
In today’s fast-changing business environment, it’s extremely important to be able to respond to client needs in the most effective and timely manner. If your customers wish to see your business online and have instant access to your products or services.
Online Grocery Store is an e-commerce website, which retails various grocery products. This project allows viewing various products available enables registered users to purchase desired products instantly using Paytm, UPI payment processor (Instant Pay) and also can place order by using Cash on Delivery (Pay Later) option. This project provides an easy access to Administrators and Managers to view orders placed using Pay Later and Instant Pay options.
In order to develop an e-commerce website, a number of Technologies must be studied and understood. These include multi-tiered architecture, server and client-side scripting techniques, implementation technologies, programming language (such as PHP, HTML, CSS, JavaScript) and MySQL relational databases. This is a project with the objective to develop a basic website where a consumer is provided with a shopping cart website and also to know about the technologies used to develop such a website.
This document will discuss each of the underlying technologies to create and implement an e- commerce website.
Hybrid optimization of pumped hydro system and solar- Engr. Abdul-Azeez.pdffxintegritypublishin
Advancements in technology unveil a myriad of electrical and electronic breakthroughs geared towards efficiently harnessing limited resources to meet human energy demands. The optimization of hybrid solar PV panels and pumped hydro energy supply systems plays a pivotal role in utilizing natural resources effectively. This initiative not only benefits humanity but also fosters environmental sustainability. The study investigated the design optimization of these hybrid systems, focusing on understanding solar radiation patterns, identifying geographical influences on solar radiation, formulating a mathematical model for system optimization, and determining the optimal configuration of PV panels and pumped hydro storage. Through a comparative analysis approach and eight weeks of data collection, the study addressed key research questions related to solar radiation patterns and optimal system design. The findings highlighted regions with heightened solar radiation levels, showcasing substantial potential for power generation and emphasizing the system's efficiency. Optimizing system design significantly boosted power generation, promoted renewable energy utilization, and enhanced energy storage capacity. The study underscored the benefits of optimizing hybrid solar PV panels and pumped hydro energy supply systems for sustainable energy usage. Optimizing the design of solar PV panels and pumped hydro energy supply systems as examined across diverse climatic conditions in a developing country, not only enhances power generation but also improves the integration of renewable energy sources and boosts energy storage capacities, particularly beneficial for less economically prosperous regions. Additionally, the study provides valuable insights for advancing energy research in economically viable areas. Recommendations included conducting site-specific assessments, utilizing advanced modeling tools, implementing regular maintenance protocols, and enhancing communication among system components.
About
Indigenized remote control interface card suitable for MAFI system CCR equipment. Compatible for IDM8000 CCR. Backplane mounted serial and TCP/Ethernet communication module for CCR remote access. IDM 8000 CCR remote control on serial and TCP protocol.
• Remote control: Parallel or serial interface.
• Compatible with MAFI CCR system.
• Compatible with IDM8000 CCR.
• Compatible with Backplane mount serial communication.
• Compatible with commercial and Defence aviation CCR system.
• Remote control system for accessing CCR and allied system over serial or TCP.
• Indigenized local Support/presence in India.
• Easy in configuration using DIP switches.
Technical Specifications
Indigenized remote control interface card suitable for MAFI system CCR equipment. Compatible for IDM8000 CCR. Backplane mounted serial and TCP/Ethernet communication module for CCR remote access. IDM 8000 CCR remote control on serial and TCP protocol.
Key Features
Indigenized remote control interface card suitable for MAFI system CCR equipment. Compatible for IDM8000 CCR. Backplane mounted serial and TCP/Ethernet communication module for CCR remote access. IDM 8000 CCR remote control on serial and TCP protocol.
• Remote control: Parallel or serial interface
• Compatible with MAFI CCR system
• Copatiable with IDM8000 CCR
• Compatible with Backplane mount serial communication.
• Compatible with commercial and Defence aviation CCR system.
• Remote control system for accessing CCR and allied system over serial or TCP.
• Indigenized local Support/presence in India.
Application
• Remote control: Parallel or serial interface.
• Compatible with MAFI CCR system.
• Compatible with IDM8000 CCR.
• Compatible with Backplane mount serial communication.
• Compatible with commercial and Defence aviation CCR system.
• Remote control system for accessing CCR and allied system over serial or TCP.
• Indigenized local Support/presence in India.
• Easy in configuration using DIP switches.
CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptxR&R Consult
CFD analysis is incredibly effective at solving mysteries and improving the performance of complex systems!
Here's a great example: At a large natural gas-fired power plant, where they use waste heat to generate steam and energy, they were puzzled that their boiler wasn't producing as much steam as expected.
R&R and Tetra Engineering Group Inc. were asked to solve the issue with reduced steam production.
An inspection had shown that a significant amount of hot flue gas was bypassing the boiler tubes, where the heat was supposed to be transferred.
R&R Consult conducted a CFD analysis, which revealed that 6.3% of the flue gas was bypassing the boiler tubes without transferring heat. The analysis also showed that the flue gas was instead being directed along the sides of the boiler and between the modules that were supposed to capture the heat. This was the cause of the reduced performance.
Based on our results, Tetra Engineering installed covering plates to reduce the bypass flow. This improved the boiler's performance and increased electricity production.
It is always satisfying when we can help solve complex challenges like this. Do your systems also need a check-up or optimization? Give us a call!
Work done in cooperation with James Malloy and David Moelling from Tetra Engineering.
More examples of our work https://www.r-r-consult.dk/en/cases-en/
Industrial Training at Shahjalal Fertilizer Company Limited (SFCL)MdTanvirMahtab2
This presentation is about the working procedure of Shahjalal Fertilizer Company Limited (SFCL). A Govt. owned Company of Bangladesh Chemical Industries Corporation under Ministry of Industries.
1. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
DOI :10.5121/ijscai.2018.7301 1
MODEL OF MULTIPLE ARTIFICIAL NEURAL
NETWORKS ORIENTED ON SALES PREDICTION AND
PRODUCT SHELF DESIGN
Alessandro Massaro, Valeria Vitti, Angelo Galiano
Dyrecta Lab, IT Research Laboratory, via Vescovo Simplicio, n. 45, 70014 Conversano
(BA), Italy
ABSTRACT
In this paper the authors proposed different Multilayer Perceptron Models (MLP) of artificial neural
networks (ANN) suitable for visual merchandising in Global Distribution (GDO) applications involving
supermarket product facing. The models are related to the prediction of different attributes concerning
mainly shelf product allocation applying times series forecasting approach. The study highlights the range
validity of the sales prediction by analysing different products allocated on a testing shelf. The paper shows
the correct procedures able to analyse most guaranteed results, by describing how test and train datasets
can be processed. The prediction results are useful in order to design monthly a planogram by taking into
account the shelf allocations, the general sales trend, and the promotion activities. The preliminary
correlation analysis provided an innovative key reading of the predicted outputs. The testing has been
performed by Weka and RapidMiner tools able to predict by MLP ANN each attribute of the experimental
dataset. Finally it is formulated an innovative hybrid model which combines Weka prediction outputs as
input of the MLP ANN RapidMiner algorithm. This implementation allows to use an artificial testing
dataset useful when experimental datasets are composed by few data, thus accelerating the self-learning
process of the model. The proposed study is developed within a framework of an industry project.
KEYWORDS
Visual Merchandising, Artificial Neural Networks, Times Series Forecasting, Weka, RapidMiner, Product
Facing, Sales Prediction, Multiple ANN, Hybrid ANN Model.
1. INTRODUCTION
Visual Merchandising (VM) is characterized in literature by the following research topics [1]:
• perspective analysis of the of the shelf;
• relationship between the various parts of the displayed products;
• movement effect induced by allocation of products;
• harmony;
• colors and lights.
Other studies have shown how the visual aspect is relevant for the consumer choice about
products allocated on a shelf [2].
2. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
2
The concept of VM is extended in literature by the study of the online visual merchandising
(OVM) [3]-[4], by confirming that the analyses performed into a store are different if
compared with e-commerce ones. For this reason the visual impact of a supermarket shelf
could be a specific research topic of VM. Another interesting research topic correlated to
VM is the customer clustering by data mining enabling business processes [5]. Also
neurosciences have been applied in the field of VM, by studying sensory effects of the
consumer in neuromarketing [6] which follows precisely the logic of artificial neural
networks (ANNs). These ANNs have also been implemented in the marketing sector for
forecasting and predictive analysis [7]. Recent studies have instead validated the neural
network approach for forecasting of stock market price, showing how such networks can be a
powerful tool for financial analysis [8]. Other studies on neural networks are in analysis of
market shares [9]. These studies suggest the use ANNs also for VM applications. Some input
variables useful for VM analysis are marketing information, environmental information [10],
and customer information [11]. In the Global Distribution (GDO) field, ANNs have been
applied in [12] to predict sales of fresh food products such as fresh milk. Other input variables
useful for the shelf-space allocation analysis are product geometry, product orientation in the
shelves [13], sales number, price, promotion percentage, shelf distance [14], and social trend
[15]. In this direction data mining can support the shelf space allocation by applying different
approaches [16]. Recent scientific papers have discussed the superior predictive properties of
particular types of neural networks such as Deep Neural Networks (DNN) which are
characterized by different data processing layers [17]. Other recent studies have shown the
applicability of ANNs to the consumer model by means of the processing of customer loyalty
and customer satisfaction variables [18]. Other researchers highlighted the importance the
accuracy estimation about the reliability of the model [19]. Finally, in [20] researchers
focused their attention on consumer behavior which can affect sales. Concerning GDO
market basket analysis (MBA) represents a good tool for shelf planogram design and in
general for business intelligence of supermarkets [21]. This last study suggests as in the
proposed research to process data monthly thus orienting the planogram design on a medium
period.
Following the main topics analyzed in state of the art will be implemented different ANN
models able to process different attributes of supermarket orienting the research outputs on
the product facing and planogram design. The proposed paper is structured as follow:
• description of the main specifications of the industry project;
• description of the dataset model and of the data pre-processing phase by explaining the
attribute meaning and their importance in the analysis;
• data processing of different times series forecasting ANN Weka and RapidMiner
models by interpreting different prediction outputs;
• definition of new hybrid model ANN Weka/RapidMiner models useful for the self-
learning phase.
3. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
3
Model 1
Model 2
Model n
Input 1
Input 2
Input 3
Input m
Output
Processing
Color legend
Top position
Shelf layout
Center position
Down position
Products in
promotion
Biscuits
Drinks
House hold
products
Planogram
Figure 1. Main functional diagram of the project workflow: planograms constructed by multiple ANN
outputs.
2. PROJECT GOAL: SYSTEM ARCHITECTURE
The industry project where the research has carried out concerns the development of different
models based on the analysis of different ANN outputs oriented on the design of shelf layouts.
In Fig. 1 is illustrated the architecture of the proposed project defined by the following input and
output specifications:
• m number of inputs variables to process (attributes);
• n number of ANN models where each model is able to predict a specific attribute of the
input dataset;
• output processing module: "combiner" of the multiple ANN outputs;
• planogram design based on the output reading.
In the first phase of the project have been defined the input data of the neural network models, by
taking into account the information found in the state of the art.
The data processing phase has been structured as follows:
4. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
4
1. data entry study by adopting appropriate databases as data sources (csv, Excel, MySQL,
big data system, etc.);
2. data pre-processing: data pre-processing as choice of fundamental attributes and as data
normalization;
3. use of other data mining algorithms useful for output interpretation (classification,
clustering, correlation Matrix, association rules, MBA indicators, etc.);
4. development of workflows and scripts related to neural network algorithms (possible use
of tools such as Rapid Miner, Orange Canvas, KNIME, Weka or others);
5. optimization of ANN parameters;
6. execution of the ANN models;
7. analysis of the accuracy of the models;
8. output reporting;
9. combination of model outputs and definition of procedures for planogram planning.
The main goal of the project is the definition of procedures suitable to plan a well-structured
planogram by combining the results of each individual neural network model. In this paper will
be focused the attention on neural network results and on output interpretations representing a
part of whole research project
3. DATASET MODEL AND PRE-PROCESSING
Below are indicated the attributes of the dataset model used for the neural network processing.
• Mese (last 47 months of sales);
• Azienda (name of the analysed GDO store);
• Incassi (monthly cash of the store);
• N_basso (number of products sold monthly which are positioned down the shelf);
• N_alto (number of products sold monthly which are positioned at the top of the shelf);
• N_Mezzo (number of products sold monthly which are positioned in the central area of the
shelf);
• Incassi (cash of monthly sales of all the products of a store);
• Incassi_scaffale (cash of monthly sales of products allocated on the analysed shelf);
• Scontrini (number of receipts pertaining the sale of the products of the whole store);
• N_promotion (product number monthly sold in promotions related the analysed shelf);
5. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
5
• N_casalinghi (number of household products monthly sold related the analysed shelf);
• N_bibite (number of drinks monthly sold related the analysed shelf);
• N_biscotti (number of biscuits monthly sold related the analysed shelf).
Some attributes are reported in the colour legend of Fig.1. In Fig. 2 is illustrated the experimental
dataset.
Figure 2. Experimental dataset made by 47 records loaded into RapidMiner local repository.
A first analysis is performed a pre-processing by plotting the product sold in the different shelf
position. As illustrated in Fig. 3 it is evident that the products allocated in the central part of the
shelf are the most sold while the products placed in the top of the shelf are the less sold.
6. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
6
Product sold placed in the shelf down position (N_basso)
Product sold placed in the shelf center position (N_Mezzo)
Product sold placed in the shelf top position (N_alto)
Months
Number
of
products
sold
Figure 3. Analysis of the products sold in the last 47 months related to different shelf layout positions.
These graphical results highlights the importance of the central position in the VM strategy. For
the pre-processing has been applied the correlation data mining algorithm by means of the
“Correlation Matrix” module of RapidMiner tool shown in Fig. 4. A correlation is a number
between -1 and +1 that measures the degree of association between two Attributes (call them X
and Y). A positive value for the correlation implies a positive association. In this case large values
of X tend to be associated with large values of Y and small values of X tend to be associated with
small values of Y. A negative value for the correlation implies a negative or inverse association.
In this case large values of X tend to be associated with small values of Y and vice versa. The
outputs of the Correlation Matrix algorithms is illustrated in the table of Fig. 4. By means of the
data correlation pre-processing analysis it is possible to observe that there is an high positive
correlation between the total number of receipts of the store and the receipts associated with the
analyzed shelf: this proves that the visual impact of the analyzed shelf provided good results. The
table of Fig. 4 indicates also that there is a significant correlation (0.3) between the number of
shelf drinks sold and receipts. Another important correlation (0.258) is checked between
N_biscotti and N_Mezzo attributes: this means, for example, that the biggest sale of biscuits
happens when products are placed in the center of the shelf. Finally it is observed a slight
correlation between N_Mezzo and N_promozione attributes, thus showing that the products
available at the center are those usually in promotions.
7. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
7
Figure 4. (Above) “Correlation Matrix” module of RapidMiner. (Below) correlation matrix results.
4. WEKA NEURAL NETWORK MODEL DEVELOPMENT AND
RESULTS
The models performed in this section are implemented in java language by using Eclipse
platform, and ANN Weka libraries. The models are based on time series forecasting with
Multilayer Perceptron -MLP- classifier [22]-[23]. A MLP represents a class of feedforward ANN,
consisting of at least three layers of nodes implementing a supervised learning technique called
backpropagation for training. Except for the input nodes, each node is a neuron that uses a
nonlinear activation function. All the predictive models proposed in this section have been
performed by an algorithm proposed in [22] which exhibits good performances and matching
with experimental values [22].
The first model is related to the prediction of cash related to a single GDO store. In Fig. 5 are
illustrated the monthly cash values and the first 6 predicted cash values (up to the 47th month are
data measured, from the 48th to 53th are the predicted values).
As example we report below a part of the main class java script concerning the attribute selections
of “Incassi” (cash) and “Mese” (month):
Istances sales = new Istances (trainreader);
// new forecaster
WekaForecaster forecaster = new WekaForecaster();
// cluster and classifier setting
forecaster.setFieldsToForecast(“Incassi”);
MultilayerPerceptron mlp = new MultilayerPerceptron();
forecaster.setBaseForecaster(mlp);
// time stamp setting
forecaster.getTSLagMaker().setTimeStampField(“Mese”);
…..
// prediction number
8. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
8
int steps =6;
Observing value amplitudes of Fig. 4 it is possible to note that only the first three value matches
with possible certain values. The other three predicted samples could provide only an uncertain
predicted values (anomalous amplitudes if compared with store cash monthly values).
Monthly cash of the store (Incassi)
Cash prediction (in future months)
Months
More certain ppredicted values
Uncertain predicted values
Cash
[Euro]
Figure 5. Weka model 1: prediction results.
The model 2 focus the attention on the cash related to a single analyzed shelf. In this case is
considered the forecast prediction of 24 months, obtaining the results of Fig. 6. From Fig. 6 it is
observed a regular trend behavior of cash that continues approximately for each 12 months. For
this case only the first four predictive results are more certain.
Uncertain predicted values
More certain ppredicted values
Months
Monthly cash shelf (Incassi_scaffale)
Cash shelf prediction (in future months)
Cash
[Euro]
Figure 6. Weka model 2: prediction results.
9. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
9
In the model 3 of Fig. 7 are illustrated the output results of the MLP ANN algorithm applied to
the receipt number prediction: as for the previous model also in this case is observed a regular
trend, and only the first five predictive results are more certain.
Months
Receipts
number
Uncertain predicted values
More certain ppredicted values
Monthly number of receipts
Prediction of number of receipts (in future months)
Figure 7. Weka model 3: prediction results.
In the model 4 of Fig. 8 are illustrated the predictive sales pertaining to the products positioned at
the top of the analyzed shelf. In this case it is observed a certain convergence of the solution for
the first 4 predicted months.
Months
Uncertain predicted values
More certain ppredicted values
Monthly number of products sold (placed on the shelf top)
Prediction of number of product sold (in future months)
Number
of
products
sold
allocated
on
the
shelf
top
Figure 8. Weka model 4: prediction results.
10. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
10
By the model 5 it is estimated the sales prediction of products placed on the shelf down part: the
first predicted four values of the irregular trend enhances an initial increase of the quantity of
products followed by a fast decrease.
Uncertain predicted values
More certain ppredicted values
Monthly number of products sold (placed on the shelf down)
Prediction of number of product sold (in future months)
Months
Number
of
products
sold
allocated
on
the
shelf
down
Figure 9. Weka model 5: prediction results.
The predictive results of model 6 shows the prediction of the product that will be sold if placed on
the middle part of the shelf, where only the first two predicted values are trusted.
Uncertain predicted values
More certain ppredicted values
Monthly number of products sold (placed on the shelf middle)
Prediction of number of product sold (in future months)
Number
of
products
sold
allocated
on
the
shelf
middle
Months
Figure 10. Weka model 6: prediction results.
Finally model 7 (Fig. 11), model 8 (Fig. 12) and model 9 (Fig. 13) indicate the sales prediction of
drinks, of household products and of biscuits placed in the middle of the shelf, respectively.
11. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
11
Uncertain predicted values
More certain ppredicted values
Months
Number
of
drinks
sold
allocated
on
the
shelf
middle Monthly number of drinks sold (placed on the shelf middle)
Prediction of number of product sold (in future months)
Figure 11. Weka model 7: prediction results.
Uncertain predicted values
More certain ppredicted values
Months
Number
of
household
products
allocated
on
the
shelf
middle
Monthly number of household products sold (placed on the shelf middle)
Prediction of number of product sold (in future months)
Figure 12. Weka model 8: prediction results.
12. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
12
Uncertain predicted values
More certain ppredicted values
Months
Number
of
biscuits
sold
allocated
on
the
shelf
middle
Figure 13. Weka model 9: prediction results.
For all the plots of Weka results have been applied the RapidMiner Scatter Multiple function, and
the “Append” object able to plot the predicted results (red lines) after the real ones (blue lines).
All the models have been executed by setting the following best parameters: learning rate of 0.3,
momentum of 0.2, training time of 500, validation set size equals to zero, seed equals to zero, and
hidden layers equals to (number of attributes + classes) / 2.
In order to estimate the variability of the results have been executed different predictions by
changing the learn rate parameter L, observing that the learning rate of L= 0.3 represents a good
compromise between convergent solutions and low variability (see Fig. 14). We observe that the
learning rate is one of the most important hyper-parameters to tune for training deep neural
networks [24]. The definition of the its optimum value will increase the prediction accuracy [24].
13. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
13
Months
Cash
shelf
prediction
Best Learning Rate
Figure 14. Weka: choice of the best learning rate indicating a slower oscillation response versus months.
All the first predicted values of the proposed Weka predictive models indicate the more certain
values. In table 1 are summarized these first values representing with green color the increase
trend and with the red color the decrease one.
Attribute Last value (47
th month)
First predicted value
(48 th month)
Increase/ decrease
trend
Cash 39209 48166
Shelf cash 1050 1325
Receipts 722 758
Down shelf 38 19
Top shelf 20 29
Middle shelf 67 39
Promotion 7 25
Household 8 14
Drinks 10 14
Biscuits 11 17
Table 1: increase/decrease trend.
The results of table 1 show that only the two attributes indicated with the red color will suffer a
decrease in sales. Let's now define some useful observations for planogram design based on the
results displayed during the simulations. For a better forecasting and shelf planning will be
considered only the initial trend. The products placed in down and middle part of the shelf will be
characterized by a decrease: being the promotion correlated with products allocated in down shelf
region (see table of Fig. 4), a corrective action could be to promote these products. Also the
products placed in middle could be promoted. The products positioned at the top, as they are
expected to increase their sales, could be repositioned in the last position. The last position could
also be maintained for household products, biscuits and drinks. Another solution could be to
move some biscuits or drinks or household products in the middle to leave the space to other
products. In any case, a greater turnout of customers and a greater total cash-out are expected, so
14. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
14
that promotional activities could be applied in a non-strong way. Furthermore, even if the current
shelf configuration is maintained, there will be a relevant increase of shelf cash, so it is not
recommended even to make large movements.
5. RAPID MINER NEURAL NETWORK MODEL : AN
ALTERNATIVE MLP ANN MODEL
MLP ANN developed by RapidMiner workflow are applied in industrial research [25]. The use of
objects constituting a workflow facilitates the execution of the MLP ANN algorithms simply
setting parameters by graphical user interfaces – GUIs-. In Fig. 15 is illustrated the RapidMiner
workflow applied for sales prediction. The developed model has been performed on the ANN
network layout reported in Fig. 16 having a single hidden layer configuration.
Figure 15. RapidMiner workflow implementing MLP ANN.
Figure 16. RapidMiner: MLP ANN layout.
15. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
15
The RapidMiner MLP ANN model is applied on cash prediction in order to propose an alternative
model, by providing the results of Fig. 17 with a good performance: an absolute error of only
(62.925 Euro +/- 52.898 Euro is checked).
Months
Monthly cash shelf (Incassi_scaffale)
Cash shelf prediction (in future months)
Cash
[Euro]
Figure 17. RapidMiner: comparison between real values of the last monthly cash shelf and predicted ones.
The model of Fig. 15 has been applied also for the prediction of the drink sales by achieving an
absolute error of 1.253 +/- 1.074 (see Fig. 18). These data processing models prove that
RapidMiner could be a good alternative to apply the models previously discussed.
Months
number
of
drinks
monthly
sold
related
the
analyzed
shelf
Monthly number of drinks sold (placed on the shelf middle)
Prediction of number of product sold (in future months)
Figure 18. RapidMiner: comparison between real values of the last monthly cash shelf and predicted ones.
16. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
16
By observing the last real values in blue color of Fig. 17 and Fig. 18, and the first ones of the
prediction indicated by the red color, it is observed a good matching with results indicated in table
1: in both case an initial increase is checked.
6. INNOVATIVE RAPIDMINER AND WEKA HYBRID MODEL
The performance of the MLP ANN algorithms could be optimized by constructing an hybrid
model involving Weka and RapidMiner algorithms. This approach is useful for cases
characterized by an experimental dataset made by few records thus allowing to accelerate the self
learning process. In Fig. 19 is illustrated the innovative hybrid model designed as a merge of
Weka and RapidMiner MLP ANN algorithms: the initial training and testing dataset are the input
data of the Weka models described in section IV, besides the RapidMiner workflow will process
the initial training dataset and, as testing, will process the output of the Weka algorithm.
RapidMiner
MLP ANN
Training
dataset
Weka MLP
ANN
Testing dataset
Output: new
testing dataset
Predictive
results
(Output 2)
Output1
(b)
I1
I2
I1
I2
Figure 19. Innovative flow chart: (a) hybrid Weka/RapidMiner model optimizing experimental dataset, (b)
RapidMiner workflow implementing the hybrid model.
By executing the flowchart of Fig. 19 (a) and Fig. 19 (b) for the prediction of drinks placed on the
shelf middle, are obtained the outputs results of Fig. 20, where output 1 indicates the outputs of
the Weka algorithm, and output 2 are the final predicted values of the RapidMiner workflow
model of Fig. 19 (b). We observe that also using testing dataset having a strong oscillation, the
final output results are characterized by a lower amplitude of oscillation.
17. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
17
Output 2
Output 1
Number of Months
Prediction
of
number
of
drinks
sold
allocated
on
the
shelf
middle
1 45
1 24
Figure 20. Hybrid model outputs: output of the Weka predictive model (output 1), and output of the final
values predicted by the RapidMiner algorithm (output 1). Months 1 refers to the first predicted values.
The comparison results of Fig. 21 highlights that the output of proposed hybrid model is
convergent to the previous one shown in Fig. 15, thus confirming that it is possible to increase the
initial experimental dataset in order to accelerate the self learning process by adding an
“artificial” testing dataset which is an output of another prediction MLP ANN model.
Months
Prediction
of
number
of
drinks
sold
allocated
on
the
shelf
middle
RapidMiner model of Fig. 15
RapidMiner hybrid model of Fig. 19
Figure 21. Hybrid model output: comparison of the outputs of the hybrid model of Fig. 19, with the outputs
of the model of Fig. 15.
18. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
18
7. CONCLUSION
The goal of the paper is the formulation of a model composed by a multiple MLP ANN networks
applied in GDO sales prediction. Each MLP ANN network is able to predict sales by analysing a
specific attribute. The experimentation is focused on the analysis of products allocated on a
supermarket shelf by providing solutions for planogram design. The prediction results are related
of total sales of a store and of sales related a testing shelf, by differentiating the analysis of some
attributes such as products placed on the top, middle and down part of the shelf, products in
promotions, and different kind of product typologies. A preliminary analysis showed some
important relationships and correlations between attributes useful for the outputs interpretation
and for the planogram design. Weka MLP ANN algorithms have been adopted for the prediction
of the initial trend of the products allocated on the testing shelf. RapidMiner MLP ANN models
have been indicated as alternative performing approaches. Finally has been proposed an hybrid
model integrating both WEKA and RapidMiner algorithms implemented in this paper. This last
model is suitable for self learning approaches processing a dataset of few sample.
ACKNOWLEDGEMENTS
The work has been developed in the frameworks of the Italian projects: “Modelli di Visual
Merchandising mediante Reti Neurali: “Neural Net Visual Merchandising” [Neural Networks
Models of Visual Merchandising: “Neural Net Visual Merchandising”]. Authors gratefully thanks
the researchers D. Barbuzzi, G. Birardi, V. Calati, L. D’Alessandro, F. De Carlo, A. Leogrande,
O. Rizzo, M. Solazzo, and M. M. Sorbo for their support in the realization of this work.
REFERENCES
[1] Bastow-Shoop, H. E., Zetocha, D. F. & Passewitz, G. (1991) "Visual Merchandising- A guide for
Small Retailers", Editor: North Central Regional Center for Rural Development, Iowa State University
of Science and Technology.
[2] Krafft, M., & Mantrala, M. K. (2010) "Retailing in the 21st Century- Current and Future trends-,"
Springer book.
[3] Katrandjiev, H. & Velinov, I. (2015) "Optimization of Online Visual Merchandising Elements
(OVME) on the Basis of Consumer Preferences," Economic Alternatives, Vol. 2, pp76-88.
[4] Lee, J., Hoch, R., Podlaseck, M., Schonberg, E. & Gomory, S. (2000) "Analysis and Visualization of
Metrics for Online Merchandising", Chapter, Web Usage Analysis and User Profiling, Volume 1836 of
the series Lecture Notes in Computer Science, pp126-141.
[5] Rajagopal, S. (2011) "Customer Data Clustering Using Data Mining Technique", International Journal
of Database Management Systems (IJDMS), Vol.3, No.4, pp1-11.
[6] Horská, E., Mehl, H. & Berík, J., (2014) "Review of Classical and Neuroscience Insights on Visual
Merchandising Elements and Store Atmosphere", Proceedings from IX. International Conference on
Applied Business Research ICABR 2014, pp284-290.
[7] Tao, J., Teng, N., Poh, H.-L. & Tan, C. L. (1998) "Forecasting and Analysis of Marketing Data Using
Neural Network", Journal of Information Science and Engineering, Vol. 14, No. 4, pp523-545.
[8] Di Persio, L. & Honchar, O. (2016) "Artificial Neural Networks Approach to the Forecast of Stock
Market Price Movements", International Journal of Economics and Management Systems, Vol. 1,
pp158-162.
[9] Khan, Z. H. ,Alin, T. S. & Hussain, A. (2011) "Price Prediction of Share Market using Artificial Neural
Network (ANN)", International Journal of Computer Applications (0975 - 8887), Vol. 22, No.2, pp42-
47.
[10]Yao, J., Teng, N., Poh, H.-L. & Tan, C. L. (1998) "Forecasting and Analysis of Marketing Data Using
Neural Networks", Journal of Information Science and Engineering, Vol. 14, No. 4, pp523-545.
[11]Rygielski, C., Wang, J.-C. & Yen, D. C. (2002) "Data Mining Techniques for Customer Relationship
Management", Technology in Society, Vol. 24, No.4, pp483-502.
19. International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI), Vol.7, No.3, August 2018
19
[12]Doganis, P., Alexandridis, A., Patrinos, P., Sarimveis, H. (2006) "Time Series Sales Forecasting for
Short Shelf-Life Food Products Based on Artificial Neural Networks and Evolutionary Computing",
Journal of Food Engineering, Vol.75, No.2, pp196-204.
[13]Murray, C. C. (2010) "Joint Optimization of Product Price, Display Orientation and Shelf-Space
Allocation in Retail Category Management," Journal of Retailing, Vol. 86, No. 2, pp125-136.
[14]Van Nierop, E., Fok, D. & Franses, P. H. (2008) "Interaction Between Shelf Layout and Marketing
Effectiveness and its Impact on Optimizing Shelf Arrangements", Marketing Science, Vol. 27, No. 6,
pp1065-1082.
[15]Kotler, P. & Zaltman, G. (1971) "Social Marketing: An Approach to Planned Social Change", Journal
of Marketing, Vol. 35, No. 3, pp3-12.
[16]Chen, M.-C. & Lin, C.-P. (2007) "Data Mining Approach to Product Assortment and Shelf Space
Allocation", Expert Systems with Applications, Vol. 32, No. 4, pp976-986.
[17]Dixon, M., Klabjan, D. & Bang, J. H. (2016)"Classification-Based Financial Markets Prediction using
Deep Neural Networks", Algorithmic Finance.
[18]Ansari, A. & Riasi, A. (2016) "Modelling and Evaluating Customer Loyalty Using Neural Networks:
Evidence from Startup Insurance Companies", Future Business Journal, Vol.2, No.1, pp15-30.
[19]Navale, G. S., Dudhwala, N., Jadhav, K., Gabda, P. & Vihangam, B. K. (2016) "Prediction of Stock
Market using Data Mining and Artificial Intelligence", International Journal of Computer Applications,
Vol. 134, No.12, pp9-11.
[20]Borimnejad, V. & Samani, R. E. (2016) "Modeling Consumer's Behavior for Packed Vegetable in
'Mayadin Management Organization of Tehran' Using Artificial Neural Network", Cogent Business &
Management, Vol. 3, pp1-14.
[21]Massaro, A., Galiano, A., Barbuzzi, D., Pellicani, L., Birardi, G., Romagno, D. D. & Frulli, L. (2017)
"Joint Activities of Market Basket Analysis and Product Facing for Business Intelligence Oriented on
Global Distribution Market: Examples of Data Mining Applications", (IJCSIT) International Journal of
Computer Science and Information Technologies, Vol. 8, No.2, pp178-183.
[22]Massaro, A., Galiano, A., Meuli, G. & Massari, S. F. (2018) "Overview and Application of Enabling
Technologies Oriented on Energy Routing Monitoring, on Network Installation and on Predictive
Maintenance", International Journal of Artificial Intelligence and Applications (IJAIA), Vol. 9, No. 2,
pp1-20.
[23]Kumar, R. & Balara, P. (2014) "Time Series Forecasting of Nifty Stock Market Using Weka", JRPS
International Journal for Research Publication & Seminar, Vol. 5, No. 2, pp1-6.
[24]Peace, I. C., Uzoma, A. O. & Ita A. A. (2015) "Effect of Learning Rate on Artificial Neural Network in
Machine Learning", International Journal of Engineering Research & Technology, Vol. 4, No. 2,
pp359-363.
[25]Massaro, A. & Galiano, A. (2018) "Design and Development of a Mobile APP for Multichannel Sales
Systems based on Artificial Intelligence", (IJCSIT) International Journal of Computer Science and
Information Technologies, Vol. 9, No. 3, pp63-68.
Corresponding Author
Alessandro Massaro: Research & Development Chief of Dyrecta Lab s.r.l.