Proceedings of the SMART–2020, IEEE Conference ID: 50582

9th International Conference on System Modeling & Advancement in Research Trends, 4th–5th, December, 2020
Faculty of Engineering & Computing Sciences, Teerthanker Mahaveer University, Moradabad, India

Bank Loan Prediction System using

Machine Learning
2020 9th International Conference System Modeling and Advancement in Research Trends (SMART) | 978-1-7281-8908-6/20/$31.00 ©2020 IEEE | DOI: 10.1109/SMART50582.2020.9336801

Anshika Gupta1, Vinay Pant2, Sudhanshu Kumar3 and Pravesh Kumar Bansal4
1,2,3
Faculty of Engineering and Computing Sciences, iNurture, TMU, Moradabad
4
Department of Computer Science, Government Engineering College, Bharatpur
E-mail: [email protected], [email protected],
3
[email protected], [email protected]

Abstract—With the advancement in technology, there This system allows jumping on particular applications that
are so many enhancements in the banking sector also. The deserve to be approved on a priority basis.
number of applications is increasing every day for loan There are some features for the prediction like-
approval. There are some bank policies that they have to ‘Gender’, ‘Married’, ‘Dependents’, ‘Education’, ‘Self_
consider while selecting an applicant for loan approval. Based
Employed’, ‘ApplicantIncome’, ‘CoapplicantIncome’,
on some parameters, the bank has to decide which one is best
for approval. It is tough and risky to check out manually every ‘LoanAmount’, ‘Loan_Amount_Term’, ‘Credit_History’,
person and then recommended for loan approval. In this ‘Property_Area’, ‘Loan_Status’.
work, we use a machine learning technique that will predict II. Literature Survey
the person who is reliable for a loan, based on the previous
record of the person whom the loan amount is accredited A prediction is a statement about what someone thinks
before. This work’s primary objective is to predict whether will happen in the future. People make predictions all the
the loan approval to a specific individual is safe or not. time. Some are very serious and are based on scientific
Keyword: Loan Dataset, Logistic Regression, Random calculations, but many are just guesses. Prediction helps us
Forest, Django. in many things to guess what will happen after some time or
I. Introduction after a year or after ten years.
As the data are increasing daily due to digitization in Predictive analytics is a branch of advanced analytics
the banking sector, people want to apply for loans through that uses many techniques from data mining, statistics,
the internet. Artificial intelligence (AI), as a typical method modeling, machine learning, and artificial intelligence
for information investigation, has gotten more consideration to analyze current data to make predictions. “Adyan Nur
increasingly. Individuals of various businesses are utilizing Alfiyatin, Hilman Taufiq [2] and their friends work on the
AI calculations to take care of the issues dependent on house price prediction. They use regression analysis and
their industry information. Banks are facing a significant Particle Swarm Optimization (PSO) to predict house price”.
problem in the approval of the loan. Daily there are so many One other similar work on the Mohamed El Mohadab,
applications that are challenging to manage by the bank Belaid Bouikhalene [3] and Said Safi to predict the rank for
employees, and also the chances of some mistakes are high. scientific research paper using supervised learning. Kumar
Most banks earn profit from the loan, but it is risky to choose Arun, Garg Ishan and Kaur Sanmeet [1] work on bank loan
deserving customers from the number of applications. One prediction on how to bank approve a loan. They proposed
mistake can make a massive loss to a bank. a model with the help of SVM and Neural networks like
Loan distribution is the primary business of almost machine learning algorithms.
every bank. This project aims to provide a loan [1, 8] to a This literature review helps us carry out our work and
deserving applicant out of all applicants. An efficient and propose a reliable bank loan prediction model.
non-biased system that reduces the bank’s time employs
checking every applicant on a priority basis. The bank III. Proposed Methodology
authorities complete all other customer’s other formalities The process to predict the bank loan of the applicants is
on time, which positively impacts the customers. The best as shown in figure 1. There is a different phase in each step,
part is that it is efficient for both banks and applicants. which is described here.

Copyright © IEEE–2020 ISBN: 978-1-7281-8908-6 423

Authorized licensed use limited to: California State University Fresno. Downloaded on June 21,2021 at 00:54:54 UTC from IEEE Xplore. Restrictions apply.
9th International Conference on System Modeling & Advancement in Research Trends, 4th–5th December, 2020
Faculty of Engineering & Computing Sciences, Teerthanker Mahaveer University, Moradabad, India

There are total 981 records of the applicants with the

values of their concerning attributes in categorical and
numerical data. The total count value of the attribute is
also shown in figure 3. In the pre-processing and feature
engineering of the data, we handle the missing value and
also normalize the data so we can further process it into
ML algorithm. The dataset is further divided into training
and testing. The model is trained on machine learning
algorithms and predicts the system on test data which is
discussed in the Next section in details.

Fig. 1: Proposed Methodology

IV. Dataset Descriptions and Pre-processing

The bank loan prediction system dataset is taken from Fig. 3: Value Count of the Parameters in the Dataset
kaggle competition which belong to different age group and
gender of the applicants. There are thirteen attributes in the A. Machine Learning
data set, such as education, married status, income, assets, Predictive analytics [10] is used to predict the data
etc. as shown in figure 2. about future events. It includes many techniques such
as data mining, machine learning [4, 9] and modeling.
Machine learning is a type of artificial intelligence
that allows a software application to learn from the
data & become more accurate in predicting outcomes
without human intervention. Machine learning and
deep learning help to design and develop such a
machine that automatically learns and predicts your
data and situation. Machine learning is often divided
into different subcategories according to the type of
problems being comes. Some ML type is as follows:
1) Supervised Learning
Supervised learning is the point at which the model
is getting prepared on a labelled dataset. In this kind of
learning both training and testing, datasets are labelled. The
output of prediction is always coming either 1 (yes) or 0
(No).
2) Unsupervised Learning
In unsupervised learning, the input data are not labelled
Fig. 2: Data Describtion and also do not have any prior information about the data.
424 Copyright © IEEE–2020 ISBN: 978-1-7281-8908-6

Authorized licensed use limited to: California State University Fresno. Downloaded on June 21,2021 at 00:54:54 UTC from IEEE Xplore. Restrictions apply.
 Bank Loan Prediction System using Machine Learning

Here the task of the machine is to find the hidden pattern Heat map is a data visualization technique that shows
from the data by using cluster analysis. The dataset is the magnitude of a phenomenon as color in two dimensions
labelled so that here we used supervised learning approach. as shown in figure 4. Color intensity shows the relationship
In our work, we used a supervised learning approach. between each other. The color variation may be by hue or
intensity, giving an obvious visual to the reader about how
B. Algorithms used for Prediction
the phenomenon is clustered or varies over space. From this
1) Logistic Regression heat map, it is negative relation in Loan amount term with
It is a classification set of rules used to assign application come attribute.
observations to a discrete set of instructions. Logistic V. Experiment and Result Analysis
regression is also a predictive analysis, like other regression
In this section, we use a machine learning algorithm on
analyses methods. Logistic regression is basically used for
a loan prediction dataset and deploy the result using HTML,
define the relationship between dependent binary variable
CSS, Django at the local server. Figure 5 shows the loan
and nominal or other independent variable. Now a day’s
prediction system of the applicants based on the value enter
logistic regression is used in many research areas like
by the bank employee.
medical science, machine learning and social science. It
also used by many e-commerce applications to predict the
mind set of customer to buy the product.
2) Random Forest
Random Forest is a robust system learning algorithm
that is used for a ramification of responsibilities along
with classification and regression. Random forests method
overcome the over fitting issue of decision trees during
training. It is an ensemble method made up of a large
number of small decision trees [5,7] called estimators where
each tree produces the prediction. The random forest model
combines the predictions of the estimators to produce a
more accurate prediction.
C. Correlation between Parameters
Fig. 5: Final Layout
The first attribute is to select the gender of the applicant
either male or female. The second is the marital status then
dependents attributes mean that the applicant is dependent
financially on someone or not. Other attributes are the
education of the customer, employment status, applicant
income, loan amount term, credit history, residential area,
etc. of the applicant. Finally, it shows the status of the loan
i.e. it is safe or risky as shown in fig 6.

Fig. 4: Heat Map Fig. 6: Final Result

Copyright © IEEE–2020 ISBN: 978-1-7281-8908-6 425

Authorized licensed use limited to: California State University Fresno. Downloaded on June 21,2021 at 00:54:54 UTC from IEEE Xplore. Restrictions apply.
9th International Conference on System Modeling & Advancement in Research Trends, 4th–5th December, 2020
Faculty of Engineering & Computing Sciences, Teerthanker Mahaveer University, Moradabad, India

VI. Conclusion and Future Scope References

[1] Kumar Arun, Garg Ishan, Kaur Sanmeer, Loan Approval Prediction
Today’s fast-growing IT industry needs to discover based on Machine Learning Approach.
new technology and update the old technology that [2] Adyan Nur Alfiyatin, Hilman Taufiq, Ruth Ema Febrita, Wayan
helps us to reduce human intervention and increase Firdaus Mahmudy, ‘Modeling House Price Prediction using
Regression Analysis and Particle Swarm Optimization’: International
the efficiency of the work. This model is used for the Journal of Advanced Computer Science and Applications (Vol. 8,
banking system or anyone who wants to apply for a No. 10, 2017).
[3] Mohamed El Mohadab, Belaid Bouikhalene, Said Safi, ‘Predicting
loan. It will be very helpful in bank management. From rank for scientific research papers using supervised learning’Applied
the analysis of the data, it is very clear that it reduces Computing and Informatics 15 (2019) 182–190.
[4] K. Hanumantha Rao, G. Srinivas, A. Damodhar, M. Vikas Krishna:
all the frauds done at the time of loan approval. Time is Implementation of Anomaly Detection Technique Using Machine
also very precious for everyone through this not only Learning Algorithms: Internatinal Journal of Computer Science and
Telecommunications (Volume2, Issue3, June 2011).
the bank but also the waiting time of the applicant will [5] J. R. Quinlan. Induction of Decision Tree. Machine Learning, Vol. 1,
also reduce. As it seems, it will not deal with some No. 1. pp. 81-106., 1086.
special cases when only one parameter is enough for [6] G. Arutjothi, C. Senthamarai: Prediction of loan status in commercial
bank using machine learning classifier, International Conference on
the decision, but it is quite efficient and reliable in Intelligent Sustainable Systems (ICISS), 2017.
some instant. [7] J.R. Quinlan. Induction of decision trees. Machine learning Springer,
1(1):81–106, 1086.
In the future, this prediction module can be more [8] Vishnu Vardhan case study of bank loan prediction,https://medium.
improved and integrated. The system is prepared on the com/@vishnumbaprof/case-study-loan-prediction-ac035f3ec9e4.
previous training data but in the future, it is possible to [9] S.S. Keerthi and E.G. Gilbert. Convergence of a generalize SMO
algorithm for SVM classifier design. Machine Learning, Springer,
make changes to software, which can accept new testing
46(1):351–360, 2002.
data and should also take part in training data and predict [10] J.M. Chambers. Computational methods for data analysis. Applied
accordingly. Statistics, Wiley, 1(2):1–10, 1077.

426 Copyright © IEEE–2020 ISBN: 978-1-7281-8908-6

Authorized licensed use limited to: California State University Fresno. Downloaded on June 21,2021 at 00:54:54 UTC from IEEE Xplore. Restrictions apply.