VISVESVARAYA TECHNOLOGICAL UNIVERSITY, BELAGAVI-590 014
SRI TARALABALU JAGADGURU INSTITUTE OF TECHNOLOGY
RANEBENNUR-581115
DEPARTMENT OF INFORMATION SCIENCE & ENGINEERING
2023-2024
Project Associates
GANESH RODDANAVAR 2SR20IS003
NAGANAGOUDA S PATIL 2SR20IS007
VENU GOPAL A 2SR20IS014
UNDER THE GUIDANCE AND CO-ORDINATION OF Head of the Department
Dr . SAVITA CHANNAGOUDAR(M Tech & PHD)
Asst. Prof., Dept. of IS&E
Dr. S G MAKANUR(M Tech & PHD)
CONTENTS
• ABSTRACT
• INTRODUCTION
• PROBLEM STATEMENT
• EXISTING AND PROPOSED SYSTEM
• OBJECTIVES
• LITERATURE SURVEY
• FUNCTIONAL REQUIREMENTS & NON-FUNCTIONAL REQUIREMENTS
• HARDWARE REQUIREMENTS & SOFTWARE REQUIREMENTS
• SYSTEM DESIGN
• DEVELOPMENT DETAILS
• IMPLEMENTATION
• TESTING
• RESULTS AND SNAPSHOT
• CONCLUSION
ABSTRACT
The increasing complexity of financial markets and the growing volume of loan transactions
necessitate effective tools for risk management.
It addresses the challenge of predicting loan default prices using machine learning
[Link] prediction of loan default prices is crucial for financial institutions to
make informed decisions, mitigate risks, and maintain a stable lending portfolio.
Various machine learning algorithms are employed to analyze historical loan data and
identify patterns that can predict the likelihood of loan default.
1
INTRODUCTION
To predict whether or not a borrower would default on a loan, financial institutions use
machine learning algorithms to analyze past loan data.
Lenders are better able to manage risk, make prudent judgments, and limit their
financial exposure to defaults if they have a good idea of when defaults will occur.
Lenders may improve decision-making, reduce exposure to risk, and secure the long-
term viability.
2
PROBLEM STATEMENT
The financial industry faces a critical challenge in accurately predicting loan default
prices to effectively manage risks associated with lending.
EXISTING SYSTEM
There is no automated facility .
Limitations of existing system
Agents or apps information may be insufficient or misleading.
3
PROPOSED SYSTEM
It is a combination of machine learning and web technology to develop a useful
product as it analyzes various attributes of the data and helps to take decisions.
Advantages of proposed system
[Link]
[Link]
[Link] Reduction
[Link] warning system
[Link] Analysis
4
OBJECTIVES
Risk Assessment
Minimize Losses
Improve Decision-Making
Enhance Efficiency
Customer Satisfaction
5
LITERATURE SURVEY
[Link]: "Predicting Credit Card Defaults Using Machine Learning
Techniques"
Author: Wei et al.
Year: 2009
This study compared different types of ML algorithms, like neural
networks and decision tree, for predicting credit card defaults. The
authors found that ensemble methods, such as random forests, were
the most effective for this task.
6
LITERATURE SURVEY contd..
[Link] : "A Comparative Study of Machine Learning Methods for
Loan-Default Prediction"
Author: Brown & Thomas
Year :2011
This study compared different types of ML algorithms, including
support vector machines, decision trees and neural networks, for
predicting loan defaults. The authors found that gradient boosting
and random forests performed the best.
7
LITERATURE SURVEY contd..
3. Title : "Predicting Loan Default: An Analysis of Variables and
Techniques"
Author : Adeyemo and Adeleke
Year : 2021
This study investigated the factors that contribute to loan default and
compared various machine learning algorithms for predicting
defaults. The authors found that gradient boosting and random
forests performed the best, and that variables such as income, loan
amount, and loan term were significant predictors of default.
8
FUNCTIONAL REQUIREMENTS
Data Collection and Preparation
Feature engineering
Model Selection
Model Validation
NON-FUNCTIONAL REQUIREMENTS
Correctness
Reliability
Maintainability
Portability
9
HARDWARE AND SOFTWARE REQUIREMENTS
HARDWARE REQUIREMENTS SOFTWARE REQUIREMENTS
System with Windows OS. NumPy
Minimum 6GB RAM. Pandas
Internet connection as per reqt. Matplotlib
Scikit -learn
Seaborn.
K Nearest Neighbours algorithm
10
SYSTEM DESIGN
SYSTEM ARCHITECTURE
11
SEQUENCE DIAGRAM
12
DATAFLOW DIAGRAM
13
DEVELOPMENT DETAILS
K Nearest Neighbors- Both classification and regression predictive issues can be solved
using KNN. It is, however, more frequently employed in classification issues in the sector .
Algorithm flow chart 14
IMPLEMENTATION
from flask import Flask, request, render_template
from joblib import load
app = Flask(_name_)
model= load('[Link]')
trans=load('transsformer')
@[Link]('/[Link]')
def homepage():
return render_template('[Link]')
@[Link]('/')
def home():
return render_template('[Link]')
@[Link]('/y_predict',methods=['POST'])
def y_predict():
'''
For rendering results on HTML GUI
'''
15
x_test = [[x for x in [Link]()]]
print(x_test)
x_test[0][0]=int(x_test[0][0])
x_test[0][2]=int(x_test[0][2])
x_test[0][6]=int(x_test[0][6])
x_test[0][7]=int(x_test[0][7])
test=[Link](x_test)
print(test)
prediction = [Link](test)
print(prediction)
if prediction[0]==1:
output='Good'
else:
output='Bad'
return render_template( '[Link]',prediction_text='Financial Risk: {}'.format(output))
@[Link]('/more_info',methods=['GET'])
def more_info():
return render_template('[Link]')
if _name_ == "_main_":
[Link](debug=True) 16
TESTING
Software testing is a very important element of the quality assurance and it
represents the SRS, designing, coding and implementation of the system
proposed.
Levels of testing:
• Test Planning
• Test Development
Types of testing:
• Unit testing
• Integration testing
• System testing
17
TEST CASES:
Admin login
Input Result Conclusion
Login type: chose Either admin is Admin has to give an
from list allowed to access the valid user name and
[Link] name : user system or an error password in order to
input message is displayed access the system, if
[Link] : user either user name or
input password is wrong an
error message is
displayed
18
Prediction
Input Result Conclusion
1. Age : user Record should be Application uses
input analyzed using pre KNN model to predict
2. Gender : user build training model the result
input and result should be
3. House : user displayed on front
input end
4. Credit amount:
user input
5. Duration : User
input
6. Account type :
user input
19
RESULTS AND SNAPSHOT
Login page
20
Web page for financial risk prediction
Entering the input values
21
Predicted output
22
CONCLUSION
Our system demonstrates the effectiveness of machine learning algorithms in
predicting loan defaults, providing valuable insights for financial institutions to
enhance risk management practices.
23
THANK
YOU!!!