Goal of this Project
Predict Ransomware based on file properties extracted from a tool. This model is a part of Full Antivirus + Malware Protection Software.
Its a classification problem (Supervised Machine Learning). The data was immbalanced and needed to be transformed (Synthetic Samples: SMOTE-Tomek).
Highlights
- LazyPredict for AutoML Official Documentation
- LIME for Local Explainations
- Weight of Evidence (Feature Selection Technique on Feature Separation Power) Read More
- Removing Multi-colinear features (VIF)
 |
|---|
| LIME Explainability for Local Interpretation |
Model Performance on Test Dataset
 |
|---|
| Confusion Matrix |
Metrics
- Model Used: Random Forest
- F1 Score: 0.99
- Matthews Correlation Coefficient (MCC): 0.985
- AUC-ROC: 0.993
Install Libraries using requirements.txt
pip install -r /path/to/requirements.txt