Sadek Harite
Data Scientist - MBA in Business Analytics, Carleton, Canada
New to Contra
Sadek is ready for their next project!
Side-by-side confusion matrix comparison across 5 ML models for fraud detection on highly imbalanced data. Models include Logistic Regression, LOF, Isolation Forest, Random Forest and XGBoost. XGBoost achieved 85 true fraud detections with only 13 false negatives.
1
20
Precision-Recall and F1 threshold analysis for a stacked ensemble fraud detection model. Optimal threshold identified at 0.977 achieving F1=0.857, Precision=0.929, Recall=0.796.
0
6
Comparison of 6 ML models on fraud detection including Logistic Regression, Random Forest, XGBoost and Stacked Ensemble. Stacked Ensemble achieved AUC of 0.978 - best performing model across all metrics
1
19
End-to-end fraud detection system using Python, XGBoost, Random Forest, SMOTE and SHAP. Achieved MCC of 0.810 and AUC of 0.978 on the ULB dataset.
0
32