Ilkogretim Online - Elementary Education Online, 2021; Vol 20 (Issue 6): pp.

1172-1180
http://ilkogretim-online.org
doi: 10.17051/ilkonline.2021.06.124

An Automated Prediction Model For College Admission

System

Dr. Arunakumari B. N* , Department of Computer Science, BMS Institute of Technology

and Management, Bengaluru, India, Email: arunakumaribn@bmsit.in

Vishnu Sastry H K , Department of Computer Science, BMS Institute of Technology and

Management, Bengaluru, India, Email: 1by18cs190@bmsit.in

Sheetal Neeraj , Department of Computer Science, BMS Institute of Technology and

Management, Bengaluru, India, Email: 1by18cs152@bmsit.in

Shashidhar R, Department of Electronics and Communication Engineering, JSS Science

and Technology University, Mysuru, Karnataka, 570006, INDIA, shashidhar.r@sjce.ac.in

Abstract.
“Life is a matter of choices, and every choice you make makes you”- John C Maxwell. At present,
many students make mistakes in their preference list of colleges because of various reasons like
inaccurate analysis of colleges, lack of knowledge, and apprehensive prediction. Later, they end
up regretting the same after allotment. Our application addresses this issue of the student
admission community. The application uses data mining and data analysis techniques. Rank,
category, preferred branches, preferred district, and preferred colleges are taken as input and
the preference list, on thorough analysis of the last five years’ cut-off data is generated. In this
paper, an attempt has been made to develop an automated web application prediction model for
a college admission system which can be used to make a wise choice of college before allotment.

Keywords: Allotment, Analysis, Data mining, Data Analysis, Cutoff, Preference List

INTRODUCTION
Objectives of the paper are as follows.
• To help students to fill their preferences at the time of option-entry process accurately.
• To ease of making better choices of college before allotment.
• To deploy a web application for college admission system.
After intermediate, students desiring to pursue engineering face lot of problem in choosing a
good college and branch of their choice. Admission into engineering colleges across states in
India happens generally through Common Entrance Tests (CET). The examination authority of
every state carries out the admission, through a centralized admission process. This admission
process happens through many rounds, depending on availability of seats. First, the students
must get their documents verified by the authority. Later, the authority releases the cut-offs of
every college, branch-wise and category- wise. Students will be allowed to give their preference
list of colleges and branches, which is also known as the option-entry process. Then, based on
rank, category and preference list given by the students, college and branch will be allotted to
them by the authority.
In each state, there are around 1 lakh seats available in nearly 300 colleges and over 35
different branches of engineering. Depending on the category, the percentage of seats in
1172 | Dr. Arunakumari B. N* An Automated Prediction Model for College Admission
System
colleges varies. There are nearly 15 different categories and hence it becomes difficult for
students to understand in which college and branch they are likely to get admitted in, even after
thorough analysis of cut-off data of the last few years. This problem becomes more serious in
case of students from reserved categories. Many students make mistakes in their preference list
or during option- entry process due to lack of knowledge, improper and incorrect analysis of
last five years’ cut-off data. Hence, such students end up not getting their deserved seat and
later regret for the same. Our idea will help in solving this problem of the student community.
This computer-aided method will minimize the stress on students, and they will be able to get
the preference list of all colleges in which they would get an admission, at the click of a button.
In this paper, we have done the necessary research using K-CET (Karnataka Common Entrance
Test) data. Our system is developed taking K-CET into consideration. Similarly, this system can
be used for Common Entrance Tests of other states and for other national level entrance exams
by just changing database used.

RELATED WORK
In the au-courant methodology, there has been considerable use of automated approaches in
education business process. These approaches can be distinguished in artificial intelligence and
conventional approaches. Several approaches of multivariate study are characteristics of the
conventional approaches; however, the adept schemes technique is a typical characteristic of
the artificial intelligence approaches. Together were adapted in several apps in the
teaching/school business like admission prediction [1] [2]. Moreover, using machine learning
and predictive modelling student admission has been predicted with high degree of accuracy.
Here, the method is not particular to the institution. Data provided by the applicant in the form
excel file containing huge records hence process needs further exploration for predicting the
student admission [3].
In [4], authors developed method that will support the organization to examine the present
scenario of student admission by anticipating the registration behaviour of student. It imparts
an approach like APRIORI examines the admission behaviour of the student by considering the
branch of the student and the branches he chosen to seek entry. The method also presents a
data-mining method naïve-bayes procedure which anticipate to which course the student can
register. Since, the student’s choices would be taken into rumination, the institution will be able
to upsurge the admission of branch based on the anticipated outcomes. And in [5], authors have
developed a web-based application system for college admission system in which students can
register their marks along with their personal info. With this application the entrance seat
allotment becomes easier and effective. However, the web-based system created with PHP has
more difficult than python.
In addition, college admission is predicted through support vector machines and perceptrons
supervised learning techniques which are based on historical applicant data [6]. This method
identifies with appropriate correctness of the eligible applicants to enroll at the institution by
the admission office based on historical data. However, this paper utilized admittance details
from a small-scale college with about 2500 students. And four prior academic years of details to
be collected to produce a sufficient sample size to create optimistic predictions.
Moreover, researchers have developed a general dataset of 41359 institution applications to
predict four-year bachelor’s graduation in a generalizable method [7] [8]. This method includes
features such as college graduation rates, sociodemographics, test scores, work skills, academic
attainment, participation in extramural events and evaluations by instructors. Still, the method
has constraints. First, though the proportions of the surveyed data cluster are extensive, it only
characterizes 41359 out of a possible 278201 applicants in the comprehensive illustration [8]
[9]. Second, illustration only comprised students those applied to colleges that recognized the
common app, recommending the possible for acceptance bias. In addition, study is essential to
examine in what way the outcomes generalize to other institutions. Third, the au-courant

1173| Dr. Arunakumari B. N An Automated Prediction Model for College Admission

System
modelling method was chosen to improve prediction accurateness. Though, it is primarily
association in nature, thus restricting explication.
Furthermore, to addressing these hindrances, there are also numerous gaps. For example, all
extant approaches focused on four-year outcome in early study, that can be increased to 6-year
outcome along with additional valued outcomes. And it would be advantageous to find
cohesions in patterns of mis-categorize and use these insights for better accurateness of
machine learning techniques.
In [10], the authors have developed a data analytics model can be used by colleges and schools
to enhance student admission. In this paper, authors have developed analytical model for a local
university based on historical built on neural networks, decision trees and logistic regression.
However, this model cannot be self-determined and only assists to compliment university
administrators' decision-making process to manage admissions and enrolment.
In [11], a hybrid recommender system has been developed for university admission system
based on knowledge recover and data mining techniques for tackling college admission
prediction problems. But the prediction model was developed for specific university and not
appropriate for other university with huge intake student admission system. In addition,
researchers have tried to develop student admission predictor system using KNN & logic
regression [12] [13]. It will aid the students to find the probabilities of their application to a
college being selected. Correspondingly, it will help students in categorizing the colleges which
are superlative fit for their profile and offer them with the particulars of those colleges. The
drawback of this research is the model considered only few universities with different rankings.
There is a need for the system to add more data related to additional colleges and disciplines
and the system can be advanced to a web-based application. Therefore, in our research an
attempt has been made to addresses this issue of the student admission community. Also, our
developed web application helps the user make wise choice of colleges for his/her option-entry.

PROPOSED METHODOLOGY
It In our proposed prediction method, we have used python machine learning libraries viz.,
pandas and numpy. And to develop the user interface (UI) and web application we have used
streamlit package. Further to deploy the web application on the internet so that it is accessible
worldwide we have used Heroku. In addition, the database consists of the average of previous
five years’ rank cut off data. The cut-off database will consist of the ranks with respect to
branch, college, and category. A candidate will obtain a rough idea regarding the seat he or she
is likely to get depending on his or her rank and category. Cut-off will be different for each
college, course, and category. The row headings consist of college names along with branches.
The column headings consist of the various categories. The data contained in the database is of
string data type. Each cell (corresponding to a branch and college i.e., row heading and category
i.e., column heading) in the database, consists of the rank that a candidate belonging to a
particular category has to secure in order to get admission into that particular branch and
college. Rank (mandatory), category (mandatory), preferred branches (optional), preferred
colleges (optional), preferred districts/location (optional), of the user are taken as input for our
method. Category is an alphanumeric value which can be selected from the drop-down list.
Preferred branches, preferred colleges and preferred districts are text inputs which are multiple
selection type and can be selected from drop-down list. Rank and category are required fields
and it is compulsory for the user to input these fields. Preferred branches, preferred colleges
and preferred districts are optional fields, and they can be used as filters for the generated
preference list and the process of generating preference entry list is as sown in figure 1.
Computational process for the proposed well-defined model is as follows.
i. Use data analytic techniques to prepare a preference list based on the user’s input. The
preference list varies depending on user input.

1174| Dr. Arunakumari B. N An Automated Prediction Model for College Admission

System
 Figure 1.Generating preference list

1175| Dr. Arunakumari B. N An Automated Prediction Model for College Admission

System
 Figure 2.Flow chart for the proposed model

ii.Retrieve data, from the database using functions from panda’s library. The functions
used are iloc, loc, query, etc.
iii.After performing the required operations, the results are stored into a new dataframe.
Later, this new data frame is sorted in ascending order of the cut-offs.
iv.Leave two blank lines after the title.

1176| Dr. Arunakumari B. N An Automated Prediction Model for College Admission

System
 Figure 3.Generate list of colleges with higher chances of obtaining a seat

Figure 4. Display chances of obtaining a seat in the preferred colleges

The proposed application offers three major functionalities:

i. Generate list of colleges with higher chances of obtaining a seat.
This functionality displays a data-frame with a list consisting of branch, college name,
location, and cut-off rank for the preferred branches in the colleges where the
candidate has high chances of obtaining a seat, in the inputted category. If no preferred
branches are selected, all the branches in the college where the candidate has high
chances of obtaining a seat are displayed as shown in figure 3.
ii. Display chances of obtaining a seat in the preferred colleges.
The user may prefer to get a seat in some college. So, the user can input preferred
college/colleges. Based on the input given, his/her chances (low/high) of getting a seat in the
preferred colleges will be displayed branch-wise according to the preferred branches as
shown in figure 4. It will also display the difference between the obtained rank and cutoff.

1177| Dr. Arunakumari B. N An Automated Prediction Model for College Admission

System
 Figure 5. Final option entry or preference list

iii. Generating option-entry or preference list

This functionality displays a data-frame containing the final option entry list consisting of
branch, college name, location, and cut-off rank for the branch in that college, in the
inputted category. The list also contains colleges where chances of getting a seat are low.
The list displayed is in sorted order of cut-offs wherein, first option is with lowest chance
of obtaining a seat and last option is with highest chance of obtaining a seat as shown in
figure 5 and the final of the proposed as shown in figure 6.

Advantages of proposed model

• Students from rural background find it difficult to do the necessary analysis and
prepare a preference list. This idea will be beneficial for them.
• Students who belong to multiple categories face difficulty in analyzing cut-offs in each
of these categories and predict the best colleges they can get an admission in.
Example: A student belonging to SC category will either choose SC-R (Scheduled
Caste- Rural) or SC-G (Scheduled Cast-General), depending on whether they are from
rural background or not respectively. However, a student from SC-R has a greater
chance of getting a better college compared to a SC-G student.
• Whatsoever is the student’s rank, this application will aid them in finding the best
branch and college for his/her rank.
• The student must input his rank, category, and preferred branches. The computer-
aided system will display the list of all the colleges he/she is likely to get admitted in.
• The student can check his / her chances of getting into preferred college/colleges.
• The location of the college is displayed along with the college name, so that user can
filter colleges based on the location.
• Also, the user can input the preferred district/districts to get list of colleges located in
that district/districts. This acts as a filter.
• The output data frame can be sorted according to user requirements i.e., according to
branch, college, or location. This acts as an additional filter.
• With this system, students can very easily obtain the detailed list of colleges, branch-
wise, category-wise, and district-wise as well.
• The system greatly reduces the stress on students and helps in making right choice of
colleges.

1178| Dr. Arunakumari B. N An Automated Prediction Model for College Admission

System
 • The same system can be used for other common entrance tests by just changing the
database (cut-off data). The codebase remains the same.

Figure 6.Output of the proposed web application

CONCLUSIONS
The web application helps the user make wise choice of colleges for his/her option-entry. Also,
the user gets an outline/rough idea of the entries they can make in the option-entry process
provided by examination authority. The same application can be used for Common Entrance
Tests of other states and for other national level entrance exams by only changing the cut-off
database of that exam. Proposed application benefits for the student admission community that
accommodates the need of students to choose the best college and helps colleges too to
recognize their stand in attracting students and finer prediction implies better results for the
students.

REFERENCES

IoannisHatzilygeroudis, AnthiKaratrantou, “An Expert System with Certainty Factors for

PredictingStudent Success”, Edited Negoita M.G., Howlett R.J., Jain L.C. Springer-Verlag
Berlin Heidelberg publisher, vol. 3213, pp. 292–298, (2004).
James SMoore, “An expert system approach to graduate school admission decisions and
academic performance prediction”, Omega International Journal of Management Science,
vol. 26, issue 5, pp. 659–670, (1998).
William Eberle et al. Using Machine Learning and predictive modeling to assess admission
policies and standards, Proceedings of 9th annual symposium, The university of
Oklahoma (2013).
HeenaSabnani, Mayur More, Prashant Kudale, SurekhaJanrao, “Prediction of Student Enrolment
Using Data Mining Techniques”, Published in International Research Journal of
Engineering and Technology Vol. 05, Issue 04, (2018).
Annam MallikharjunaRoa et al., College Admission Predictor, published in Journal of Network
Communications and Emerging Technologies (JNCET), Vol. 8, Issue 4, (2018).
1179| Dr. Arunakumari B. N An Automated Prediction Model for College Admission
System
Thomas Lux et al., “Applications of Supervised Learning Techniques on Undergraduate
Admissions Data”, CF '16: Proceedings of the ACM International Conference on
Computing Frontiers, May 16-19, Pages 412–417, (2016).
S. Hut, M. Gardener, D. Kamentz, A. Duckworth, and S. D’Mello, “Prospectively Predicting 4-Year
College Graduation from Student Applications,” In International Conference on Learning
Analytics and Knowledge, March 7–9, Australia, pp. 412–417, (2018).
KanadpriyaBasu, TreenaBasu, Ron Buckmire, and NishuLal, “Predictive Models of Student
College Commitment Decisions Using Machine Learning,” vol. 4, issue 2, pp. 1-18, (2019).
Austin Waters and RistoMiikkulainen, GRADE: Machine Learning Support for Graduate
Admissions, In Proceedings of the 25th Conference on Innovative Applications of
Artificial Intelligence, Association for the Advancement of Artificial Intelligence, (2013).
Jared Cirelli, Andrea M. Konkol, Faisal Aqlan, “Predictive Analytics Models for Student
Admission and Enrolment, in proceedings of the International Conference on Industrial
Engineering and Operations Management Washington DC, USA, September 27-29,
(2018).
A. H. M. Ragab, A. F. S. Mashat and A. M. Khedra, HRSPCA: A hybrid recommender system for
predicting college admission, in proceedings of 12th international conference on
intelligent system design and applications, pg. 107-113, (2012).
HimanshuSonawane, “Student Admission Predictor”, MSc Research Project Data Analytics,
National College of Ireland Project Submission Sheet –School of Computing (2018).
Mane, R.V., Ghorpade, V.R., “Predicting student admission decisions by association rule mining
with pattern growth approach” In: 2016 International Conference on Electrical,
Electronics, Communication, Computer and Optimization Techniques pg. 202–207,
(2016).

1180| Dr. Arunakumari B. N An Automated Prediction Model for College Admission

System