This project focuses on automating loan eligibility prediction for Dream Housing Finance, a company that provides home loans across urban, semi-urban, and rural areas. The goal is to identify customer segments eligible for loan approval using machine learning models, enabling the company to target potential borrowers effectively.
The company wants to automate the loan approval process based on customer details submitted during the online application. These details include income, loan amount, credit history, education, marital status, and more. Given a dataset of past applicants, we analyze the data and build predictive models to classify whether a loan should be approved (Loan_Status = Y) or rejected (Loan_Status = N).
The dataset consists of customer details relevant to loan eligibility. The key features include:
| Variable | Description |
|---|---|
| Loan_ID | Unique identifier for each loan application |
| Gender | Applicant's gender (Male/Female) |
| Married | Marital status (Yes/No) |
| Dependents | Number of dependents |
| Education | Education level (Graduate/Undergraduate) |
| Self_Employed | Whether the applicant is self-employed (Yes/No) |
| ApplicantIncome | Applicant's monthly income |
| CoapplicantIncome | Income of co-applicant |
| LoanAmount | Loan amount requested (in thousands) |
| Loan_Amount_Term | Loan term duration (months) |
| Credit_History | Whether credit history meets guidelines (1/0) |
| Property_Area | Area of the property (Urban/Semi-Urban/Rural) |
| Loan_Status | Loan approval status (Y/N - Target Variable) |
- Handling missing values and data inconsistencies
- Visualizing relationships between features and loan status
- Analyzing trends based on income, credit history, and loan amount
- Understanding the impact of categorical variables (e.g., Property Area, Education)
- Encoding categorical variables
- Handling skewness and outliers in numerical variables
- Scaling features for model compatibility
- Creating new features (if necessary) to improve predictions
Three machine learning models were implemented:
✅ Logistic Regression – Baseline model to understand linear relationships
✅ Decision Tree – Captures non-linear patterns in loan approval trends
✅ Random Forest – An ensemble model improving accuracy and reducing overfitting
- Accuracy Score: Measures overall correctness of predictions
- Precision & Recall: Evaluates model effectiveness in detecting eligible borrowers
- Confusion Matrix: Analyzes classification errors
- ROC-AUC Curve: Assesses model performance in distinguishing approvals vs. rejections
📌 Credit History is the strongest predictor of loan approval. Applicants with a good credit history (1) have a significantly higher approval rate.
📌 Income and Loan Amount show a relationship—higher income applicants tend to receive loan approvals more easily.
📌 Marital Status & Dependents have minor impacts, but self-employed applicants face slightly higher rejection rates.
📌 Random Forest performed best, delivering the highest accuracy and robust classification.
🚀 Hyperparameter tuning to optimize Decision Tree & Random Forest models
🚀 Feature Engineering to incorporate additional financial indicators
🚀 Deploying the model via a web app for real-time loan approval predictions
🟢 Python (Pandas, NumPy, Scikit-learn, Matplotlib, Seaborn)
🟢 Jupyter Notebook for data analysis and model development
🟢 Machine Learning (Logistic Regression, Decision Tree, Random Forest)
This project successfully demonstrates the power of machine learning in loan approval prediction. By leveraging EDA, feature engineering, and predictive modeling, we provide insights into customer eligibility, helping Dream Housing Finance streamline loan processing and target high-potential borrowers efficiently.