Nihal Kumar 00290202021

Summer Training Project

Loan Eligibility Prediction using

Machine Learning

Name: Nihal Kumar

Enrollment No.: 00290202021

Semester & Section: 5A

Nihal Kumar 00290202021

PROBLEM STATEMENT
1) The process of validation and verification is time-consuming and requires a significant
amount of time and effort.

2) During the validation process, there is a possibility of introducing human errors, which can
affect the accuracy of the results.

3) There is a lack of cross-referencing previous loan records, which can lead to inconsistencies
and potential errors in the validation process.

4) The validation process requires a large number of human resources, which can be a
significant cost and time burden for the organization.

WHY THE PARTICULAR TOPIC IS CHOSEN? IT MUST ADDRESS THE

PRESENT STATE OF ART
The chosen topic for the data science and machine learning project is Loan Eligibility
Prediction. This topic is chosen because it is a critical problem faced by banks and loan
companies, and accurate prediction can help in reducing the risk of default and improving the
loan approval process. The present state of the art in this field involves the use of machine
learning algorithms and optimization techniques to develop accurate and efficient loan
eligibility prediction models. These models use various factors such as credit score, past loan
history, income, and other background information of the applicant to pr edict loan eligibility.
The use of machine learning models has shown promising results in accurately predicting loan
eligibility and reducing the risk of default. The project can contribute to the present state of the
art by developing an accurate and efficient loan eligibility prediction model using machine
learning algorithms and optimization techniques.

OBJECTIVE AND SCOPE OF THE PROJECT

The primary objective of this search is to extract patterns from a common loan-train dataset,
and then building a model which will make the accurate prediction and help banks to make
approving the loan very easy.

The historical data of customers will be used in order to do the analysis.

To make the process of loan approval easy using fewer resources.

Nihal Kumar 00290202021

ANALYSIS, DESIGN, DEVELOPMENT & TESTING METHODOLOGIES

1. Analysis Phase:

Problem Definition: Define the problem statement and objectives clearly. In this case,
the goal is to predict whether a loan applicant is eligible for a loan based on various
features.

Data Collection: Gather relevant data sources, including applicant information, financial
history, and loan approval status. Data can be collected from internal databases,
external sources, or APIs.

Data Exploration: Explore the dataset to understand its structure, quality, and
distribution. Identify missing values, outliers, and potential data issues.

Feature Engineering: Select and preprocess features that are relevant to the prediction
task. This may include encoding categorical variables, handling missing data, and scaling
numerical features.

Data Splitting: Split the dataset into training, validation, and testing sets to evaluate the
model's performance accurately.

2. Design Phase:

Model Selection: Choose appropriate machine learning algorithms for classification

tasks. Common choices include logistic regression, decision trees, random forests, and
support vector machines.

Model Architecture: Design the architecture of your machine learning model, including
the number of layers and neurons for neural networks, or the depth of decision trees.

Validation Strategy: Determine the evaluation metrics (e.g., accuracy, precision, recall,
F1-score) and validation strategy (e.g., k-fold cross-validation) to assess the model's
performance.

3. Development Phase:

Data Preprocessing: Preprocess the training data by applying the feature engineering
techniques identified during the analysis phase.
Nihal Kumar 00290202021

Model Training: Train the selected machine learning model on the training dataset.
Optimize hyperparameters using techniques like grid search or random search.

Model Evaluation: Evaluate the model's performance on the validation dataset using
the chosen metrics. Tweak the model and repeat this step until you achieve satisfactory
results.

Model Deployment: Once the model meets the desired performance criteria, deploy it
in a production environment, such as a web application or an API for loan eligibility
prediction.

4. Testing Phase:

Model Testing: Assess the model's performance on the test dataset to ensure it
generalizes well to unseen data.

Error Analysis: Analyze model errors to understand common patterns or

misclassifications. This can help in fine-tuning the model or improving the dataset.

Monitoring and Maintenance: Implement monitoring to keep track of model

performance in real-time and update the model as needed. This ensures that the model
remains accurate as the data distribution changes over time.

H/W & S/W BE USED

Hardware Used

1) Windows Computer

Software/Code Edit Used

1) Jupiter Notebook

TESTING TECHNOLOGIES TO BE USED

White-Box Testing
Nihal Kumar 00290202021

WHAT CONTRIBUTION/ VALUE ADDITION WOULD THE PROJECT MAKE?

1) Improved loan approval process: The project can help in developing an accurate and
efficient loan eligibility prediction model that can reduce the risk of default and improve the
loan approval process.

2) Identification of relevant attributes: The project can identify the most relevant attributes
that affect the prediction result the most, such as credit score, past loan history, income, and
other background information of the applicant.

3) Automation of loan eligibility process: The project can automate the loan eligibility process
by using machine learning models to predict the approval probability of each application.

4) Reduction of risk of default: The project can reduce the risk of default by accurately
predicting loan eligibility and identifying potential defaulters.

5) Contribution to the present state of the art: The project can contribute to the present state
of the art by developing an accurate and efficient loan eligibility prediction model using
machine learning algorithms and optimization techniques.

LIMITATIONS / CONSTRAINTS OF THE PROJECT

1) Availability of Data: The project requires a large amount of historical data of customers,
including their credit score, past loan history, income, and other background information. The
availability of such data can be a constraint for the project.

2) Data Quality: The quality of the data used for the project is crucial for the accuracy of the
loan eligibility prediction model. The data should be accurate, complete, and free from errors or
biases.

3) Model Complexity: The complexity of the machine learning model used for the project can
be a limitation. A complex model may require more computational resources and time to train
and may not be easily interpretable.

4) Model Overfitting: Overfitting is a common problem in machine learning models, where the
model performs well on the training data but poorly on the test data. Overfitting can be a
limitation for the project, and techniques such as regularization can be used to prevent it.

5) Ethical Considerations: The loan eligibility prediction model should be developed and used
ethically, without any discrimination or bias against any group of people. The model should
comply with the legal and ethical standards of the industry.
Nihal Kumar 00290202021

6) Interpretability: The interpretability of the loan eligibility prediction model is important for
transparency and accountability. The model should be easily interpretable, and the factors that
affect the prediction result should be understandable to the stakeholders.

7) Scalability: The loan eligibility prediction model should be scalable to handle a large volume
of loan applications and customer data. The model should be able to handle new data and
adapt to changing market conditions.

CONCLUSION AND FUTURE SCOPE FOR MODIFICATION

Conclusion:

The system approves or rejects the loan applications. Recovery of loans is a major contributing
parameter in the financial statements of a bank. It is very difficult to predict the possibility of
payment of loan by the customer. Machine Learning (ML) techniques are very useful in
predicting outcomes for large amount of data. In our project, three machine learning
algorithms, Logistic Regression (LR), Decision Tree (DT) and Random Forest (RF) are applied to
predict the loan approval of customers. The experimental results conclude that the accuracy of
Random Forest machine algorithm is better than compared to Logistic Regression and decision
tree machine learning approaches.

Future Scope for Modification:

1. Feature Engineering: Explore additional features that may have an impact on loan
eligibility, such as the applicant’s employment history, debt-to-income ratio, and loan
purpose.
2. Model Selection and Optimization: Experiment with different machine learning
algorithms and optimization techniques to find the most accurate and efficient model
for loan eligibility prediction.
3. Ensemble Learning: Combine multiple models to create an ensemble model that can
further improve the prediction accuracy.
4. Real-time Prediction: Develop a system that can provide real-time loan eligibility
predictions based on the applicant’s input and updated data.
5. Interpretability: Analyze the interpretability of the models, which can help us
understand the factors that contribute to loan eligibility prediction.
Nihal Kumar 00290202021

USE CASE

DATASET DESCRIPTION
- Loan_ID: Unique identifier for each loan applicant

- Gender: Gender of the loan applicant

- Married: Marital status of the loan applicant

- Dependents: Number of dependents of the loan applicant

- Education: Education level of the loan applicant

- Self_Employed: Whether the loan applicant is self-employed or not

Nihal Kumar 00290202021

- ApplicantIncome: Income of the loan applicant

- CoapplicantIncome: Income of the co-applicant (if any)

- LoanAmount: Loan amount applied for

- Loan_Amount_Term: Term of the loan in months

- Credit_History: Credit history of the loan applicant

- Property_Area: Area where the property is located

- Loan_Status: Whether the loan was approved or not

About Data

 What is the name of dataset file?

o loan-train.csv
 What is the format of the data?
o Data is in tabular format.
 What is the data taken from?
o Kaggle
 How large is the database (in numbers of rows and columns)?
o 501 rows × 14 columns
 What data types are present (symbolic, numeric, etc.)?
o float64(4), int64(1), object(8)
Nihal Kumar 00290202021

FLOW CHART

REFERENCE

https://www.geeksforgeeks.org/loan-eligibility-prediction-using-machine-learning-models-in-
python/

https://www.kaggle.com/code/vikasukani/loan-eligibility-prediction-machine-learning