Smart Loan Pro

Project report submitted to

Indian Institute of Information Technology, Nagpur, in
partial fulfillment of the requirements for the award of
the degree of

Bachelor of Technology
In
Computer Science and Engineering
by

Asmit Raj(BT20CSE205) Keshaw Kumar(BT20CSE100)

Mohit Anand(BT20CSE067) Yash Anand(BT20CSE186)

Under the guidance of

Dr. Khushboo Jain

Indian Institute of Information

Technology, Nagpur 441108 (India)

2020-2024
 Smart Loan Pro

Project report submitted to

Indian Institute of Information Technology, Nagpur, in
partial fulfillment of the requirements for the award of
the degree of

Bachelor of Technology
In
Computer Science and Engineering
by

Asmit Raj(BT20CSE205) Keshaw Kumar(BT20CSE100)

Mohit Anand(BT20CSE067) Yash Anand(BT19CSE186)

Under the guidance of

Dr. Khushboo Jain

Indian Institute of Information

Technology, Nagpur 441108 (India)

2024
© Indian Institute of Information Technology, Nagpur (IIIT) 2024
 Declaration

We, Asmit Raj (BT20CSE205), Keshaw Kumar (BT20CSE100), Mohit

Anand (BT20CSE067), Yash Anand (BT20CSE186) hereby declare that our
project work titled “Smart Loan Pro” is carried out by us in the Department of
Computer Science and Engineering at Indian Institute of Information Technology,
Nagpur. The work is original and has not been submitted earlier whole or in part for the
award of any degree/diploma at this or any other Institution / University.

Date: 17th May 2024

Asmit Raj Keshaw Kumar Mohit Anand Yash Anand

(BT20CSE205) (BT20CSE100) (BT20CSE067) (BT20CSE186)

 Declaration

We, Asmit Raj (BT20CSE186), Keshaw Kumar (BT20CSE100), Mohit Anand

(BT20CSE067), Yash Anand (BT20CSE186) understand that plagiarism is defined
as any one or the combination of the following:

1. Uncredited verbatim copying of individual sentences, paragraphs or

illustrations (such as graphs, diagrams, etc.) from any source, published or
unpublished, including the internet.
2. Uncredited improper paraphrasing of pages or paragraphs (changing a few
words or phrases, or rearranging the original sentence order).
3. Credited verbatim copying of a major portion of a paper (or thesis chapter)
without clear delineation of who did or wrote what. (Source: IEEE papers) I have
made sure that all the ideas, expressions, graphs, diagrams, etc. that are not a result of
my own work, are properly credited. Long phrases or sentences that had to be used
verbatim from published literature have been clearly identified using quotation marks.

I affirm that no portion of my work can be considered plagiarism, and I take full
responsibility if such a complaint occurs. I understand fully well that the guide of the
thesis may not be in a position to check for the possibility of such incidences of
plagiarism in this body of work.

Date: 17th May 2024

Asmit Raj Keshaw Kumar Mohit Anand Yash Anand

(BT20CSE205) (BT20CSE100) (BT20CSE067) (BT20CSE186)

 Certificate

This is to certify that the project titled “Smart Loan Pro”, submitted by Asmit Raj
(BT20CSE186), Keshaw Kumar (BT20CSE100), Mohit Anand (BT20CSE067),
Yash Anand (BT20CSE186) in the partial fulfillment of the requirements for the award
of the degree of Bachelor of Technology in Computer Science and Engineering, IIIT
Nagpur. The work is comprehensive, complete and fit for final evaluation.

Date: 17th May, 2024

Dr. Khushboo Jain

Project Guide, Computer Science and Engineering, IIIT Nagpur

Dr. Tausif Diwan

Head of Department, Computer Science and Engineering, IIIT Nagpur

 Acknowledgement

“Acknowledgement is an art, one can write glib stanzas without meaning a Word, and
on the other hand one can make a simple expression of gratitude”

It gives us a great sense of pleasure to present the report of the Project Work undertaken
during B. Tech. Final Year. We owe a special debt of gratitude to our Project Mentor Dr
Khushboo Jain, Department of Computer Science Engineering, Indian Institute of
Information Technology, Nagpur for his constant support and guidance throughout our
work. It is only his cognizant efforts that our endeavors have seen the light of the day.
No amount of written expression is sufficient to show our deepest scene of gratitude to
him.

We would also like to express our sincere thanks to the members of the project review
committee for their continuous support and guidance. Their invaluable insights have
allowed us to broaden our knowledge in new fields. We would also like to acknowledge
the contributions of all the faculty members of the department for their kind assistance
and cooperation during the development of our project.

We hold a deep sense of reverence and gratitude towards our parents and family
members, whose unwavering moral and financial support have been invaluable to us
throughout this journey.

Lastly, we would like to express our gratitude to all our friends who directly or indirectly
aided us in the completion of this project and the preparation of this report.

We are truly indebted to all the individuals mentioned above for their indispensable
contributions, without which this research report would not have been possible.

Asmit Raj Keshaw Kumar Mohit Anand Yash Anand

(BT20CSE205) (BT20CSE100) (BT20CSE067) (BT20CSE186)

 Abstract

Smart Loan Pro is a comprehensive full stack website designed to revolutionize the
borrowing experience by seamlessly integrating cutting-edge technology with financial
expertise. In today's fast-paced world, access to transparent and reliable loan information
is paramount for making informed financial decisions. Smart Loan Pro addresses this
need by offering a user-friendly platform that simplifies the loan application process and
provides personalized recommendations tailored to individual financial goals. In
particular, our project entailed the development of a user-centric website, providing a
seamless experience for individuals to access our loan services. Through this platform,
users are empowered to effortlessly login and submit digital loan applications,
significantly reducing administrative overhead.
The website also offers valuable resources and educational content to empower users
with financial literacy and knowledge. From informative articles on managing debt to
interactive tools for budget planning, Smart Loan Pro aims to equip users with the skills
and insights needed to make sound financial decisions both now and in the future.
Smart Loan Pro prioritizes transparency and security, ensuring that users have access to
clear and concise information about loan terms, fees, and potential risks. Robust security
measures are implemented to safeguard users' sensitive data, maintaining the highest
standards of privacy and confidentiality.

I
 Table of Contents

Serial No. Contents Page No.

1 Introduction 1
1.1 Motivation for the work 2
1.2 The objective of the work 2
1.3 Organization of thesis 2

2 Literature Review 4

2.1 Study I: Loan Prediction 4

2.2 Study II: Loan Service Applications 5

2.3 Study II: Loan Service Application 6

3 Work Done 7

3.1 Dataset 7

3.2 Website for Client 9

3.3 Admin Dashboard 17

4 Results and Discussions 21

5 Summary and Conclusions 25

II
 List of Figures

Figure No. Title Page No.

Figure-3.1.1 Chatbot using for Registration 8

Figure-3.1.2 Chatbot using for Tracking 8

Figure-3.2.1 Sign Up Form 9

Figure-3.2.2 Login Page 10

Figure-3.2.3 Apply loan button 10

Figure-3.2.4 Loan Application Form 11

Figure-3.2.5 EMI Calculator 12

Figure-3.2.6 EMI Calculator Result 12

Figure-3.2.7 Amortization Schedule 13

Figure-3.2.8 User Profile and update profile 13

Figure-3.2.9 User Loan History 14

Figure-3.2.10 Update loan history 14

Figure-3.2.11 Loan Details 15

Figure-3.2.12 Contact Us 16

Figure-3.3.1 Admin Page 16

Figure-3.3.2 All register users 17

Figure-3.3.4 User Profile view from Admin 18

Figure-3.3.5 Loan History view for Admin 18

III
 List of Tables

Table No. Title Page No.

Table 4.1 Comparison Table of Metrics 22

IV
 Chapter 1

Introduction

In the rapidly evolving world of financial technology, convenience and effectiveness

have united to create a revolutionary platform: Smart Pro Loan. Getting loans for things
like buying a house, paying for personal expenses, and attending school has often been a
challenging process when we navigate the complex waters of personal finance.
Recognising the need for empowerment and simplicity in financial decision-making,
Smart Pro Loan is a leader in innovation, skillfully fusing state-of-the-art features to
completely change the way we manage loan applications.

The present study explores the various facets of Smart Pro Loan, an all-inclusive fintech
platform that leads the way in accessibility, user experience, and predictive analytics.
This platform goes above and beyond the norm by integrating state-of-the-art tools like
an interactive loan predictor, a careful credit score calculator, and a chatbot that responds
to users' inquiries. Its main goal is to give users a one-stop shop for all of their varied
loan needs.

By including these extra-cool features, the loan application process as a whole is

improved in addition to the user interface. By examining the features of Smart Pro Loan,
this report aims to shed light on how technology can be used to expedite loan application
processes, demystify credit assessments, and empower consumers to make informed
financial decisions.By including these extra-cool features, the loan application process as
a whole is improved in addition to the user interface. By examining the features of Smart
Pro Loan, this report aims to shed light on how technology can be used to expedite loan
application processes, demystify credit assessments, and empower consumers to make
informed financial decisions.

1
1.1 Motivation for the work
This project is driven by a fundamental commitment to advancing financial
inclusion and accessibility in contemporary society. Our primary objective is
to eliminate barriers and extend loans to individuals without established
credit histories, fostering financial inclusion on a broader scale. We seek to
revolutionise the loan application process, prioritising simplicity and
modernization to accommodate the fast-paced lifestyles of today. By
providing a straightforward and reliable avenue for small firms to secure
loans, we aim to empower and bolster their growth. Utilising cutting-edge
technology such as AI chatbots and loan predictors, we strive to enhance the
overall user experience, ensuring a seamless and informative journey through
the application process. Additionally, our dedication to financial literacy is
manifested through tools like an EMI calculator, aimed at promoting
informed decision-making among users.

1.2 The objective of the work

Our project's primary objectives encompass fostering financial inclusion by

extending loans to individuals and startups overlooked by traditional financial
services due to credit history or other barriers. We aim to revolutionize the
loan application process by implementing a user-friendly online platform,
eliminating the need for physical visits to a bank and reducing paperwork.
Additionally, we aspire to enhance user experience through the integration of
an AI-powered chatbot, acting as a knowledgeable assistant for answering
queries and generating necessary loan-related documents. Furthermore, our
project seeks to deploy AI algorithms to create a loan eligibility prediction
tool, providing quick feedback on users' likelihood to qualify for a loan based
on specific parameters.

1.3 Organization of Thesis

The rest of the thesis is organized as follows:

Chapter 2
This chapter focuses on the Literature Survey of the project.

2
Chapter 3
This chapter gives a detailed report of the work completed.

Chapter 4
This chapter gives a detailed report of the work completed.

Chapter 5
This chapter provides the overall summary of the project.

Chapter 6
This chapter includes references and datasets mentioned throughout the
report.

3
 Chapter 2

Literature Review

2.1 Study I: Loan Prediction

This study employs a bank customer dataset with 7 numerical and 6 categorical
attributes to predict loan approval. Using Principal Component Analysis (PCA)
for feature reduction, the extracted features are inputted into various machine
learning algorithms, including logistic regression, SVM, and KNN. In the
Pre-Processing Stage, missing values are addressed, and One-hot encoding
transforms categorical data. The logistic regression algorithm stands out with the
highest accuracy of 81.1% on the original dataset, making it the most effective
model for predicting loan approval.

In this study on loan prediction enhancement, missing values were handled by

filling them with the mode during pre-processing. Feature selection employed
Pearson correlation, with the chosen features used as inputs for machine learning
algorithms. The dataset included 981 applicant records with categorical and
numerical attributes. The proposed model showcased strong performance, with
Logistic Regression achieving 89.80%, Random Forest classifier reaching
85.41%, and Decision Tree yielding 77.46% accuracy. These findings emphasize
the efficacy of the chosen algorithms in improving loan prediction for robust
credit assessment systems.

This study focuses on predicting loan approval in banks using a machine learning
approach with a Kaggle-sourced dataset. Feature selection via mutual information
was followed by pre-processing, involving filling missing values with column
modes and log transformation to mitigate outlier influence. The proposed model,
evaluated with Naive Bayes (83.73% accuracy), Random Forest (77.23%),
Decision Tree (63.41%), and KNN (77.46%), demonstrates its effectiveness in
loan approval prediction, with Naive Bayes yielding the highest accuracy.

4
2.2 Study II: Loan Service Application
In the assessment of automatic text summarization methods for chatbots,
TextRank emerges as the top-performing extractive algorithm, excelling in
ROUGE 1, ROUGE 2, and ROUGE L metrics. Conversely, PEGASUS
outperforms all other abstractive algorithms in the same metrics. The evaluation
utilised the BBC News dataset from Kaggle (2225 documents from 2004–2005),
and the proposed model integrates LexRank, TextRank, LSA, T5, BART, and
PEGASUS. Evaluation employed ROUGE metrics, including ROUGE 1,
ROUGE 2, ROUGE L, and ROUGEsum, providing a comprehensive analysis of
summarization efficacy for conversational chatbot applications.

The Abstractive EmailSum project aims to summarise email threads using a

dataset of 2,549 examples. The dataset is split into testing (500 threads) and
training (1,800 threads) sets, with an average document length of 233 words. The
proposed model combines extractive approaches like Oracle, Lead, TestRank,
and BertSumExt, with abstractive techniques such as Fast Abs RL, T5, Transfer
Learning, and Hierarchical T5. This comprehensive methodology addresses the
complexities of email thread summarization.

In this study on automatic text summarization of COVID-19 research articles, a

dataset of over 59,000 scholarly articles is employed. The abstractive
summarization model uses a pre-trained distill version of GPT-2, with
specifications including a token length of up to 1024, 12 attention heads, and 6
transformer decoder layers. The unsupervised extractive summarization method
involves a BERT model with 12 attention heads and 6 transformer encoder layers.
Results show that a higher compression ratio (60%) in extractive summarization
improves ROUGE scores, and abstractive summarization benefits from more
input tokens (60% extraction) for higher ROUGE scores. Using only verbs as
keywords yields low ROUGE scores, while using only nouns performs similarly
to using both verbs and nouns.

5
2.3 Study III: Loan Application
The loan service application employs a robust methodology for website
development, ensuring user-friendly interfaces and secure authentication through
Oracle Access Manager (OAM). User profiles capture essential details, including
name, age, gender, email, Aadhar card, phone number, and address. Content is
efficiently published using the Adobe Experience Manager (AEM) with the
Angular JavaScript framework. Fiserv/DNA organizes customer transaction
details, offering account summary functions like viewing detailed loan
information and scheduling online payments. The application supports various
repayment methods, including equal installments, one-time payments, and
recurring payments, with timely notifications through instant access to monthly
statements and email alerts, enhancing the overall user experience.

6
 Chapter 3

Work Done

3.1 Chatbots

In today's digital era, chatbots have become integral, transforming user

interactions through NLP, enhancing user experiences, enabling efficient
communication, and ensuring 24/7 availability.

Chatbots play a crucial role in enhancing the user experience through a range of
services. Our Chatbots provides tracking of their loan applications, applying for
loans, personalized loan information, detailing various products, features, and
eligibility criteria based on the user's profile. Additionally, our chatbots guide
users seamlessly through the loan application process, offering assistance in form
completion and document uploads, ensuring a user-friendly experience.
Frequently Asked Questions (FAQs) are intelligently handled by the chatbot,
providing users with instant and accurate responses.

Google's DialogFlow framework, a versatile tool for creating NLP Chatbots. The
development process involved creating agents, defining intents, handling default
scenarios, specifying user responses, and leveraging context for nuanced
interactions. Utilizing FastAPI as the backend facilitated the seamless connection
between the front-end chatbot interface and the underlying logic and data.

7
Fig. 3.1.1 Chatbot using for Registration

Fig. 3.1.2 Chatbot using for Tracking

8
3.2 Website for Client
Signup Page: The signup page serves as the entry point for new users to create
accounts on the platform. Ensuring a smooth user experience, we designed this
page to collect essential information including a unique email address, password,
full name, phone number, and address. Each field is carefully validated to prevent
duplicate entries and maintain data integrity. By requiring unique email addresses
and phone numbers, we minimize the risk of duplicate accounts while ensuring
that users can be easily identified and contacted.

Fig. 3.2.1 Sign Up Form

Login Page: Security is paramount in any fintech application, especially during

the authentication process. Our login page is designed to securely authenticate
users using their registered email and password. Upon successful authentication,
an access token is generated, providing secure access to the user's account and
protecting sensitive information from unauthorized access.

9
 Fig. 3.2.2 Login Page

Loan Application: Our loan application feature is meticulously crafted to

streamline the borrowing process for users. We provide a user-friendly interface
where users can input all necessary details required for loan processing. This
includes personal information, financial details, and the purpose of the loan. By
guiding users through each step of the application process and ensuring that all
required information is collected, we aim to simplify the loan application process
and expedite loan approvals.

Fig. 3.2.3 Apply loan button

10
 Fig. 3.2.4 Loan Application Form

Calculators: Financial planning tools are essential for empowering users to make
informed decisions about their finances. We have integrated several calculators
into the platform, including EMI (Equated Monthly Installment), SIP (Systematic
Investment Plan), and FD (Fixed Deposit) calculators. These calculators enable
users to calculate loan repayment schedules, plan their investments, and estimate
returns on fixed deposits, thereby promoting financial literacy and enabling users
to make sound financial decisions.

11
 Fig. 3.2.5 EMI Calculator

Fig. 3.2.6 EMI Calculator Result

Profile Management: Our profile management system allows users to update their
personal information, view their loan history, and manage their loan updates with
ease. Users can edit their profile details such as contact information and address,
ensuring that their information is always up-to-date. Additionally, users can track
the status of their loan applications and receive updates on any changes or
decisions made by the lending institution.

12
 Fig 3.2.7 Amortization Schedule

Fig 3.2.8 User Profile and update profile

13
 Fig 3.2.9 User Loan History

Fig 3.2.10 Update loan history

14
Fig 3.2.11 Loan Details

15
 Additional Pages: In addition to core functionality, we have included additional
pages such as Contact Us, About Us, and Blog sections to enhance user
engagement and provide valuable resources. The Contact Us page enables users to
reach out to customer support for assistance or inquiries. The About Us page
provides information about the company, its mission, and its values, fostering
trust and transparency. The Blog section serves as a hub for informative articles
and resources related to finance, lending, and personal finance management,
enriching the user experience and providing valuable insights.

Fig. 3.2.12 Contact Us

3.3 Admin Dashboard

Fig 3.3.1 Admin Page

User Management: The admin dashboard empowers administrators with

16
comprehensive user management capabilities. Administrators can view all users
registered on the platform, update user profiles, and delete user accounts if
necessary. This functionality ensures that administrators have full control over
user accounts and can efficiently manage user data. Additionally, administrators
can view specific user loan details, allowing them to monitor user activity and
address any issues or concerns promptly.

Fig 3.3.2 All register users

17
 Fig 3.3.3 User Profile view from Admin

Loan Management: Effective loan management is crucial for ensuring the

smooth operation of the lending platform. Our admin dashboard provides
administrators with the tools they need to manage loans efficiently.
Administrators can delete loans that do not meet the platform's criteria, update the
status of loans as they progress through the approval process, and monitor various
loan types to identify trends or patterns. By centralizing loan management tasks
within the admin dashboard, we streamline administrative processes and facilitate
effective decision-making.

Fig 3.3.4 Loan History View for Admin

18
In summary, each component of the loan fintech application is meticulously designed
and developed to deliver a seamless user experience and empower administrators with
the tools they need to efficiently manage the platform. From user registration and loan
applications to profile management and administrative controls, every aspect of the
application is crafted with the user's needs and security in mind.

3.4 Technologies/ Framework used

In our research, we have utilized a range of cutting-edge technologies, frameworks, and

libraries that are widely recognized in the machine learning and deep learning
communities.

Our primary deep learning framework is Keras, which provides a user-friendly interface
to build and train deep neural networks. Keras is built on top of TensorFlow, a popular
open-source machine learning framework developed by Google, which we have used
extensively for our experiments.

For our development environment, we have used Spyder, an open-source integrated

development environment (IDE) that provides powerful debugging and editing features
for Python code. Additionally, we have used Google Colab, a cloud-based platform that
provides free access to GPU and TPU resources for running our experiments at scale.
To deploy our machine learning models as a web application, we have used Streamlit, an
open-source Python library that makes it easy to create interactive web applications for
machine learning and data science. We have also utilized other libraries such as Pandas,
NumPy, Matplotlib, Scikit-learn, and PyTorch to preprocess data, visualize results, and
implement various machine learning algorithms.

19
 Chapter 4

Results and Discussions

The registration and login functionalities provide users with personalized access to the
platform, allowing them to securely access their accounts and perform various actions.
This feature enhances user engagement and security by ensuring that only registered
users can access the loan application process and other functionalities. Additionally, it
enables the platform to collect user data for personalized services and targeted
marketing.

Users can conveniently apply for loans online, streamlining the traditional loan
application process. The online loan application feature significantly improves user
convenience and accessibility. By eliminating the need for physical paperwork and
in-person visits to a bank or lending institution, users can apply for loans from the
comfort of their homes or anywhere with internet access. This feature also reduces
processing time and increases efficiency for both users and the lending institution.

Users are required to sign up or log in before applying for a loan, ensuring that all
applicants have authenticated identities. Requiring users to sign up or log in before
accessing the loan application process enhances security and accountability. It allows the
platform to verify the identity of applicants and maintain a record of their loan
applications and interactions. Additionally, it enables the platform to offer personalized
loan recommendations and services based on the user's profile and history.

Users have the capability to track the status of their loan applications directly from their
profiles on the Smart Loan Pro website. This feature adds transparency and convenience
to the loan application process. Users can easily monitor the progress of their
applications, including any pending documents, approval status, or additional steps
required. By providing real-time updates through the user's profile, the platform
enhances user satisfaction and reduces the need for users to contact customer support for
status inquiries.

The loan application status feature is seamlessly integrated into the user profiles,

20
ensuring easy access and navigation. By incorporating this functionality into user
profiles, the platform centralizes all relevant information and actions for users,
streamlining the user experience. Users can effortlessly switch between different sections
of their profiles to check their application status, update personal information, or perform
other account-related tasks. This integration fosters a cohesive and intuitive user
interface, enhancing overall usability and satisfaction.

Users have access to an EMI (Equated Monthly Installment) calculator tool on the Smart
Loan Pro website. This feature empowers users to estimate their monthly loan repayment
amounts based on various loan parameters such as principal amount, interest rate, and
loan tenure. By providing a user-friendly EMI calculator, the platform enhances financial
transparency and helps users make informed decisions regarding their loan
commitments. The EMI calculator allows users to input customizable parameters such as
loan amount, interest rate, and loan tenure. Offering customizable input fields enables
users to tailor the EMI calculation to their specific loan requirements and financial
circumstances. Users can experiment with different loan scenarios and adjust parameters
to understand how changes in loan terms impact their monthly repayment obligations.
This flexibility enhances user empowerment and facilitates better financial planning.

The admin dashboard provides a comprehensive view of all loan applications submitted
through the platform. Admins can access and review details of each loan application,
including applicant information, requested loan amount, and status. This centralized view
enables efficient monitoring and management of the loan pipeline, facilitating timely
processing and decision-making. Admins have the authority to approve or reject loan
applications directly from the dashboard. Admins can take decisive actions on loan
applications based on established criteria and policies. They may review applicant
qualifications, creditworthiness, and compliance with lending regulations before making
approval decisions. Rejected applications may be accompanied by reasons for the
decision, providing transparency to applicants. Admins can manage user profiles,
including viewing, editing, or updating user information. Admins have the capability to
assist users with profile-related inquiries or issues, such as updating contact details or
verifying identity documents. This feature ensures data accuracy and compliance with
regulatory requirements while offering support to users who may require assistance with
their profiles.

21
We obtained a loan dataset from Kaggle and proceeded with data preprocessing steps to
ensure the dataset's suitability for machine learning model development. Initially, we
addressed missing values in categorical features, including 'Gender', 'Dependents',
'Married', 'Education', 'Self_Employed', 'Loan_Amount_Term', and 'Credit_History', by
filling them with zeros. The missing values in the 'Loan_Amount' column were imputed
using the median value.

To facilitate model training, we encoded categorical variables such as 'Gender',

'Dependents', 'Married', 'Education', 'Self_Employed', 'Credit_History', and the target
variable 'Loan_Status' into numerical representations (0 and 1). Subsequently, we applied
one-hot encoding to the 'Property_Area' feature, which included categories such as
"Urban," "Rural," and "Semiurban."

The dataset was split into training and testing sets, adhering to an 80:20 ratio. Due to the
small size of the dataset, we addressed class imbalance by employing the Synthetic
Minority Over-sampling Technique (SMOTE) on the training set, ensuring a more
balanced representation of the target classes.

we proceeded to build and evaluate machine learning models, including K-Nearest

Neighbors (KNN), Decision Tree, Random Forest, Support Vector Machine (SVM),
Naive Bayes, Logistic Regression, Stochastic Gradient Descent, and an ensemble model.
Comparative analysis of these models was conducted based on Train F1 score, Test F1
score, and Test Accuracy metrics. Notably, the ensemble model emerged as the most
promising, exhibiting the highest Test Accuracy among the models considered.

Table 4.1 Comparison Table of Metrics

Models Train F1 Score Test F1 Score Test Accuracy

KNN 0.831753 0.908163 0.853659

Decision Tree 0.855633 0.917526 0.869919

Random Forest 0.830781 0.886486 0.829268

SVM Classifier 0.858106 0.918367 0.869919

22
 Naive Bayes Classifier 0.809149 0.894737 0.837398

Logistic Regression 0.857152 0.907216 0.853659

Stochastic Gradient 0.782861 0.918367 0.869919

Descent

ENSEMBLE 0.857243 0.918367 0.869919

Table 4.1 Comparison Table of Metrics

This comprehensive project report outlines the systematic steps taken in data
preprocessing, encoding, oversampling, feature scaling, and model building, culminating
in the identification of the most effective machine learning model for loan approval
prediction.

23
 Chapter 5

Summary and Conclusions

Our project represents a transformative leap in the lending landscape, bringing together
cutting-edge technologies to create a robust and user-centric platform. The core
components include a machine learning-driven loan prediction module, an interactive
NLP chatbot for customer engagement, and a CIBIL Score Calculator for rapid credit
assessments.

The loan prediction model harnesses the power of machine learning algorithms to provide
users with personalized insights into their likelihood of loan approval. Complementing
this, the NLP chatbot, developed using Google's framework for DialogFlow, enhances
customer interaction by offering an engaging and efficient communication channel.

Additionally, our project features a sophisticated CIBIL Score Calculator, enabling users
to quickly assess their creditworthiness. The full-stack website seamlessly integrates
these functionalities, ensuring a user-friendly experience throughout the loan application
process.

By employing a diverse tech stack and leveraging the strengths of each component, our
project streamlines and automates crucial aspects of the lending journey. The result is a
comprehensive financial ecosystem that prioritizes efficiency, user experience, and data
security, setting a new standard for innovation in the finance and technology intersection.

24
 References

[1] Mohammad Ahmad Sheikh; Amit Kumar Goel; Tapas Kumar. An Approach for
Prediction of Loan Approval using Machine Learning Algorithm. Published in: 2020
International Conference on Electronics and Sustainable Communication Systems
(ICESC), Coimbatore, India, 02-04 July 2020

[2] R Nancy Deborah; S Alwyn Rajiv; A Vinora; C Manjula Devi; S Mohammed Arif.
An Efficient Loan Approval Status Prediction Using Machine Learning. Published in:
2023 International Conference on Advanced Computing Technologies and Applications
(ICACTA), Mumbai, India, 06-07 October 2023

[3] Praveen Tumuluru; Lakshmi Ramani Burra; M. Loukya; S. Bhavana; H.M.H.

CSaiBaba; Comparative Analysis of Customer Loan Approval Prediction using Machine
Learning Algorithms. Published in: 2022 Second International Conference on Artificial
Intelligence and Smart Energy (ICAIS), Coimbatore, India, 23-25 February 2022

[4] Sk. Sharmila; P V S Sandhya; P. Suhani Kousar; P. Anuradha; S. Deekshitha, Bank

Loan Approval using Machine Learning. Published in: 2024 International Conference on
Integrated Circuits and Communication Systems (ICICACS), 23-24 February 2024,
Raichur, India

[5] Vishal Singh; Ayushman Yadav; Rajat Awasthi; Guide N. Partheeban. Prediction of
Modernized Loan Approval System Based on Machine Learning Approach. Published
in: 2021 International Conference on Intelligent Technologies (CONIT), 25-27 June
2021, Hubli, India

[6] Krishna Mridha; Dipayan Barua; Meghla Monir Shorna; Hasan Nouman Nouman.
Credit Approval Decision using Machine Learning Algorithms. Published in: 2022 10th
International Conference on Reliability, Infocom Technologies and Optimization
(Trends and Future Directions) (ICRITO), Noida, India. 13-14 October 2022

[7]. Vishal Dutt; Satya Murthy Sasubilli; Anand Eswararao Yerrapati. Dynamic
Information Retrieval With Chatbots: A Review of Artificial Intelligence Methodology.
Published in: 2020 4th International Conference on Electronics, Communication and
Aerospace Technology (ICECA) Coimbatore, India, 05-07 November 2020

[8] Szűcs, Gábor and Németh, Marcell (2021) Double-View Matching Network for
Few-Shot Learning to Classify Covid-19 in X-ray images. INFOCOMMUNICATIONS
JOURNAL, 13 (1). pp. 26-34. ISSN 2061-2079

[9] Rachana Vannala; S.B. Swathi; Yuvaraj Puranam. AI Chatbot For Answering FAQ’s.
Published in: 2022 IEEE 2nd International Conference on Sustainable Energy and
Future Electric Transportation (SeFeT). Hyderabad, India, 04-06 August 2022

[10] Sarthak Kesarwani; Titiksha; Sapna Juneja. Student Chatbot System: A Review on
Educational Chatbot. Published in: 2023 7th International Conference on Trends in
Electronics and Informatics (ICOEI), Tirunelveli, India, 11-13 April 2023

[11] Salomo Hendrian Sudjono; Favian HN Adrian; Caroline Angelina Sunarya;

Gabrielle Felicia. Comparison of Different Machine Learning Algorithms for Predicting
Loan Risk Categories, Jakarta, Indonesia, 16-16 February 2023

[12] Atharva Thodge; Harsh M; C. Pretty Diana Cyril. Web-based chatbot for basic
financial and mortgage services. Published in: 2023 2nd International Conference on
Vision Towards Emerging Trends in Communication and Networking Technologies
(ViTECoN), Vellore, India, 05-06 May 2023.

[13] Ajmeera Kiran; I. Jeya Kumar; P. Vijayakarthik; S.K Lokesh Naik; T. Vinod.
Intelligent Chat Bots: An AI Based Chat Bot For Better Banking Applications.
Published in: 2023 International Conference on Computer Communication and
Informatics (ICCCI), Coimbatore, India, 23-25 January 2023.

[14] Balaji Vasan Srinivasan; Nathan Gnanasambandam; Shi Zhao; Raj Minhas.
Domain-Specific Adaptation of a Partial Least Squares Regression Model for Loan
Defaults Prediction. Published in: 2011 IEEE 11th International Conference on Data
Mining Workshops, Vancouver, BC, Canada, 11-11 December 2011.

[15] Material UI. Online Documentation https://mui.com/material-ui/

[16] Fast Api. online Documentation https://fastapi.tiangolo.com/

[17] Python 3.12.3. Online Documentation https://docs.python.org/3/

[18] React 18.0 Online Documentation https://react.dev/