FinApprove - Loan Approval Prediction

A machine learning project that predicts whether a loan application will be approved or not, based on applicant financial and demographic data.

---

About the Project

FinApprove uses supervised classification algorithms to analyze loan application data and predict approval outcomes. The project covers the full ML pipeline — from data cleaning and exploratory analysis to model training, evaluation, and feature engineering.

---

Dataset

File: loan_approval_data.csv

The dataset includes the following features:

Feature	Description
Applicant_ID	Unique identifier for each applicant (dropped before training)
Gender	Gender of the applicant
Marital_Status	Marital status of the applicant
Education_Level	Education background
Employment_Status	Whether the applicant is employed
Employer_Category	Type of employer
Applicant_Income	Monthly income of the applicant
Coapplicant_Income	Monthly income of the co-applicant
Credit_Score	Credit score of the applicant
DTI_Ratio	Debt-to-Income ratio
Savings	Savings amount
Loan_Purpose	Purpose of the loan
Property_Area	Area type of the property
Loan_Approved	Target variable (Yes / No)

---

Project Workflow

1. Data Loading

• Loaded dataset using pandas

2. Handling Missing Values

• Numerical columns filled using mean imputation
• Categorical columns filled using most frequent value imputation

3. Exploratory Data Analysis (EDA)

• Class distribution of loan approval (pie chart)
• Gender and education level distribution (bar plots)
• Income distribution for applicant and co-applicant (histograms)
• Outlier analysis using box plots (Income, Credit Score, DTI Ratio, Savings)
• Relationship between Credit Score and Loan Approval

4. Encoding

• LabelEncoder applied to Education_Level and Loan_Approved
• OneHotEncoder (drop first) applied to: Marital_Status, Employment_Status, Loan_Purpose, Gender, Employer_Category, Property_Area

5. Correlation Analysis

• Heatmap generated to identify relationships between features
• Top correlated features with Loan_Approved identified

6. Feature Scaling

• StandardScaler applied to training and test sets

7. Model Training and Evaluation

Three classification models were trained and evaluated:

Model	Metrics
Logistic Regression	Accuracy, Precision, Recall, F1, Confusion Matrix
K-Nearest Neighbors (K=5)	Accuracy, Precision, Recall, F1, Confusion Matrix
Gaussian Naive Bayes	Accuracy, Precision, Recall, F1, Confusion Matrix

8. Feature Engineering

• Added squared features: DTI_Ratio_sq and Credit_Score_sq
• Original DTI_Ratio and Credit_Score columns dropped after engineering
• Gaussian Naive Bayes retrained on engineered features

---

Tech Stack

• Language: Python 3
• Notebook: Jupyter Notebook
• Libraries:
  • pandas — data manipulation
  • numpy — numerical operations
  • matplotlib & seaborn — data visualization
  • scikit-learn — preprocessing, model training, and evaluation

---

Installation and Setup

1. Clone the repository

git clone https://github.com/your-username/FinApprove.git
cd FinApprove

2. Install required libraries

pip install pandas numpy matplotlib seaborn scikit-learn

3. Place the dataset in the project folder

Make sure loan_approval_data.csv is in the same directory as the notebook.

4. Run the notebook

jupyter notebook FinApprove.ipynb

---

Project Structure

FinApprove/
│
├── FinApprove.ipynb          # Main Jupyter Notebook
├── loan_approval_data.csv    # Dataset (add manually)
└── README.md                 # Project documentation

---

Models Used

• Logistic Regression — baseline linear classifier
• K-Nearest Neighbors — distance-based classifier
• Gaussian Naive Bayes — probabilistic classifier, also tested with engineered features

---

License

This project is open source and available under the MIT License.