💳 Credit Risk Classifier using Decision Tree

A machine learning project to predict whether a loan applicant is a good or bad credit risk using the kaggle German Credit Dataset with Target.

---

📊 Project Objective

To build and evaluate a classification model that predicts the creditworthiness of applicants based on demographic and financial features.

---

🧠 Algorithms Used

• ✅ Decision Tree Classifier (with pre-pruning)
• ✅ GridSearchCV for hyperparameter tuning

---

📂 Dataset Overview

• Source: UCI Machine Learning Repository
• Samples: 1000
• Features: Age, Job, Housing, Saving accounts, Checking account, Credit amount, Duration, Purpose, etc.
• Target: Risk (good/bad)

---

🔧 Key Steps Performed

📌 Data Preprocessing

• Handled missing values using mode imputation for categorical fields
• Applied One-Hot Encoding for multi-class categorical features
• Scaled numerical features (Age, Credit amount, Duration) using RobustScaler

🧪 Model Training

• Trained a Decision Tree Classifier with:
  • Pre-pruning (max_depth, min_samples_split, min_samples_leaf)
  • class_weight='balanced' to handle class imbalance

📈 Evaluation

• Confusion Matrix
• Precision, Recall, F1-Score
• ROC Curve and AUC Score
• Comparative ROC plots between multiple decision tree configurations

---