Loan Prediction Machine Learning Project 📊💡

Welcome to the Loan Prediction Machine Learning Project repository! This project focuses on predicting loan eligibility based on various customer attributes. By leveraging classification algorithms, we aim to develop a robust model that accurately assesses whether a customer qualifies for a loan.

Overview 🌟

Our project encompasses the following key stages:

1. Data Exploration: We conducted an in-depth analysis of the dataset to understand the impact of attributes like marital status, education, and employment status on loan eligibility.

2. Data Preprocessing: We handled missing values, outliers, and encoded categorical variables to prepare the data for model training.

3. Feature Analysis: We explored trends and correlations in the data, uncovering insights into factors affecting loan approvals.

4. Model Implementation: We implemented and compared classification models, including Logistic Regression and K-Nearest Neighbors (KNN), to determine the best-performing algorithm for loan prediction.

5. Performance Evaluation: The models were evaluated using accuracy, precision, recall, and F1-score metrics. Logistic Regression emerged as the superior model with an accuracy of 79% compared to KNN's 77%.

6. Predicting Loan Status: We used the Logistic Regression model to predict loan status for new customers and analyzed the results to gain further insights.

Key Features 🚀

• Logistic Regression: A powerful classification model that achieved an accuracy of 79%, effectively identifying eligible and ineligible loan applicants.

• K-Nearest Neighbors (KNN): An alternative classification model evaluated using GridSearchCV to find the optimal number of neighbors. Although KNN performed well, Logistic Regression proved to be more accurate.

• Data Insights: Analyzed patterns in the new customer data to identify trends such as the percentage of married individuals in semiurban areas who obtained loans.

• Visualization: Utilized visualizations to understand the distribution of loan status across different attributes like marital status and employment.

Getting Started 🚀

1. Clone the Repository:

git clone https://github.com/yourusername/loan-prediction-project.git

2. Navigate to the Project Directory:

cd loan-prediction-project

3. Install Dependencies:

pip install -r requirements.txt

4. Run Data Analysis and Model Training Scripts:

To explore and preprocess the data:

jupyter notebook data_analysis.ipynb

To train and evaluate the models:

python train_models.py

Performance Metrics 📈

• Logistic Regression:

  • Accuracy: 79%
  • Precision:
    • Class 0: 0.95
    • Class 1: 0.76
  • Recall:
    • Class 0: 0.40
    • Class 1: 0.99
  • F1-score:
    • Class 0: 0.56
    • Class 1: 0.86

• K-Nearest Neighbors (KNN):

  • Accuracy: 77%
  • Precision:
    • Class 0: 0.81
    • Class 1: 0.76
  • Recall:
    • Class 0: 0.42
    • Class 1: 0.95
  • F1-score:
    • Class 0: 0.55
    • Class 1: 0.84

Insights and Analysis 🔍

• The Logistic Regression model demonstrates superior performance compared to K-Nearest Neighbors (KNN) with a higher accuracy of 79% versus 77%.
• Key insights from the new customer data reveal significant trends, such as the percentage of married individuals in semiurban areas who secured loans.