Loan Approval Prediction

This project focuses on predicting loan approval using machine learning algorithms. The model takes various customer features as input and predicts whether a loan application will be approved or not. Key steps in the workflow include data preprocessing, feature engineering, model selection, and hyperparameter tuning.

Project Overview

• Dataset: Includes features related to loan applicants such as income, credit history, loan amount, etc.
• Objective: Build a model to predict loan approval status.
• Methods Used: Decision Trees, GridSearchCV for hyperparameter tuning, and PCA for dimensionality reduction.

Key Features

1. Data Preprocessing:

   • Handle missing values, scaling, and encoding categorical features.

2. Modeling:

   • Use Decision Tree Classifier as the base model.
   • Perform hyperparameter tuning with GridSearchCV to optimize model performance.

3. Evaluation Metrics:

   • Accuracy Score: 0.96
   • Confusion Matrix and Classification Report show excellent performance on precision, recall, and F1-score.

Installation

1. Clone the repository:

   git clone https://github.com/alihassanml/Loan-Approval-Prediction.git

2. Install the necessary dependencies:

   pip install -r requirements.txt

Usage

1. Open the Jupyter Notebook:

   jupyter notebook Loan_Approval_Prediction.ipynb

2. Follow the steps in the notebook to preprocess data, build, and evaluate the model.

Hyperparameter Tuning

• Parameters Tuned:

  • Criterion: ['gini', 'entropy', 'log_loss']
  • Splitter: ['best', 'random']
  • Max Depth: [1, 5, 10, 15, 20, 25, 30]
  • Max Features: ['auto', 'sqrt', 'log2']

• Best parameters were selected using 5-fold cross-validation for optimal accuracy.

Results

• Accuracy Score: 0.96
• Confusion Matrix:

  [[854  33]
   [ 23 499]]
  

• Classification Report:

                precision    recall  f1-score   support
         0       0.97      0.96      0.97       887
         1       0.94      0.96      0.95       522
  

Future Work

• Explore other classification algorithms like Random Forest, SVM, or XGBoost.
• Further optimize the model with additional feature engineering techniques.