2. •To develop a website for CAT Preparation
by using the concepts of dynamic web
designing, machine learning and data
analytics, so that the user not only gets the
course material, but also analysis,
recommendations and advice based on his/her
performance.
•To compare the performance of multiple
machine learning algorithms for clustering and
classification.
3. My CAT Gym is a website that would combine the
features of traditional preparation system for CAT
and concepts of dynamic web designing, machine
learning and data analytics to become a
contemporary preparation system for CAT that not
only provides the basic functionalities like course
material and tests, but also does their analysis and
provides the user with recommendations to improve
their performance so that they can excel in the CAT
exam, which is the most values exam for admission
to any MBA college.
4. Common Admission Test for Management (CAT) is an
exam which is targeted by a majority of MBA
aspirants. With the advancement in technology, all the
course material is available on the internet. But, there
is no such website that facilitates personalized training
for the subjects covered in the exam, that not only
provides the course material, but also advisory for the
preparation.
This project also serves as a testing base for
comparison of performance of various classification
and clustering algorithms (both unsupervised and
supervised), which has also been done as a part of the
project.
5. •K – Means Clustering Algorithm using
Euclidean Distance
•K – Means Clustering Algorithm using
Manhattan Distance
•Proposed Algorithm (Range Clustering)
6. The primary objective of this project is to provide
complete guidance to the users, i. e., CAT aspirants by
using the concepts of dynamic web designing, machine
learning and data analytics.
In addition to what the traditional systems offer, the
project would be having dedicated mechanism for
categorizing analyzing a topic as the aspirants strength
or weakness. This analysis would help the aspirant plan
better by focusing more on the weaker are rather than
repeatedly working on the area which actually don’t
need that much attention.
7. •When a user opens the website as a guest user, he/she has access to only
information part but not the features part, which are only available to the
registered/ logged in users.
•When a user registers on the website, some entries are made in the progress
table in the database for monitoring the status of user’s progress in his/her
preparation. When the user opens a page, for example, topic notes of topic
like percentages, a variable in the progress table is set for that user
indicating that the user has read the topic notes of percentages. This is done
for all the users, all the topics and for the features video, notes, practice
questions and practice tests.
•When a registered user gives a test, his/her score, the correct answers and
the incorrect answers are stored in a table in the database. Now, depending
upon which algorithm has been deployed for determining the status of topic
that the user just gave a test, the topic is classified as easy, moderate or
tough for the user. This helps in providing a competitive and adaptive
preparation to the user.
10. CAT is one of the most popular exams that MBA aspirants
give. But, till date, there has not been any solution that
focuses on each aspirant individually. The package offered by
coaching institutes such as Career Launcher, TIME, MBA
Guru, Alchemist, etc. and by online resources like
www.hitbullseye.com are not providing dedicated monitoring
of user’s progress and performance.
By using the machine learning concepts, this website would
be a breakthrough out of the above mentioned traditional
preparation methods as it not only provides the course
material in the form of videos and nutshell notes, but also
personalized recommendations that help the user to plan
better.
11. On applying different clustering and classification algorithms on the
website, it has been observed that the Dynamic Average Algorithm is
computationally the least expensive (script time = 1 sec) and produces the
most efficient clusters and categorizes a topic the best. In the beginning
when the number of users was less, k – Means Algorithm was working
fine but as the number of users, and hence, the number of computations
increased, it’s performance started to decrease (from script time = 4 sec to
script time = 12 sec). Agglomerative Algorithm using Manhattan
Distance is the least efficient of all in providing clusters, having the
longest script time (>20 sec).It did not work well with the data of my
website even when the number of users was less. One reason for this
could be that it requires large random samples for best performance,
which could not be provided by the database of my website at this point
of time.
Hence, till now, the Dynamic Average Algorithm is working most
12. ALGORITHM NO. OF USERS QUESTIONS FEATURES SCRIPT TIME
K MEANS - EUCLIDEAN DISTANCE 10 7 3 ~ 8 – 15 sec
K MEANS - MANHATTAN DISTANCE 10 7 3 ~ 18 – 23 sec
PROPOSED (RANGE CLUSTERING) 10 7 3 ~ 1 sec
K MEANS - EUCLIDEAN DISTANCE 20 7 3 ~ 10 – 16 sec
K MEANS - MANHATTAN DISTANCE 20 7 3 ~ 19 – 24 sec
PROPOSED (RANGE CLUSTERING) 20 7 3 ~ 1 sec
K MEANS - EUCLIDEAN DISTANCE 20 15 3 ~ 15 – 19 sec
K MEANS - MANHATTAN DISTANCE 20 15 3 ~ 20 – 24 sec
PROPOSED (RANGE CLUSTERING) 20 15 3 ~ 1 sec
K MEANS - EUCLIDEAN DISTANCE 20 15 2 ~ 8 – 12 sec
K MEANS - MANHATTAN DISTANCE 20 15 2 ~ 18 – 23 sec
PROPOSED (RANGE CLUSTERING) 20 15 2 ~ 1 sec
13. •More efficient clustering can be done once a quality
training data set is available
•A feature of Live Streaming of lectures by experts in the
field can be added
•Question bank for practice questions as well as practice
tests can be enriched
•A social forum can be created where the registered users
can discuss their hardships
•An online library feature can be added that will have the
previous year papers of various MBA entrance exams and
other reference material (like, self help books)
•A feature of admissions counseling can be added