Overview

A machine learning model that detects fraudulent transactions with 97% accuracy using Random Forest classification

Results

• Accuracy:97 %
• Fraud Recall: 84%
• Fraud Precision: 64%

Tools Used

-Python -Scikit- learn -Smote (imbalanced-learn) -Pandas, NumPy, Matplotlib

Steps

1. Generated sythetic fraud dataset (10,000 records)
2. Handled class imbalance using SMOTE
3. Trained Random Forest (100 trees)
4. Tuned model with GridSearchCV (97.4% recall)
5. Visualized results with confusion matrix

Key Concepts Demostrated

• Class imbalance handling
• Model Tuning
• Feature importance analysis
• Classification metrics interpretation