Bank loan forecast

This project was carried out in September 2021 as part of a predictive analysis competition organised on Analytics Vidhya.
The challenge: to create a model capable of predicting whether a loan application will be accepted by a bank, based on numerous socio-economic criteria.

❓ Problematic

Financial institutions receive a large number of loan applications.
➡️ How to automate and ensure the reliability of the eligibility decision-making through machine learning, while considering factors such as income, previous credit, family status, etc.?

🛠️ Solution implemented

• 🔎 Exploratory analysis of the data to identify trends and discriminating factors

• 🧼 Cleaning and preprocessing of the data (missing values, encoding categorical variables)

• 🧠 Implementation of several models:

  • XGBoost

  • Random Forest

  • Logistic Regression

• 📈 Comparison using classification metrics: accuracy, precision, recall, ROC AUC

• 💡 Feature selection to improve performance and reduce overfitting

Link to the project GitHub repository: Loan Prediction

⚙️ Technical stack

• Language: Python

• Libraries: scikit-learn · xgboost · pandas · seaborn

• Methods: Supervised Classification · Exploratory Analysis · Data Preprocessing

• Environment: Jupyter Notebook

• Prerequisites: Machine Learning · Statistics · Tabular Data Processing