🤖 Android Malware Detection

XAI Pipeline | RandomForest | SHAP · LIME · DiCE · MITRE ATT&CK

Dataset: Drebin-215 | 15,036 samples | 215 features

⚙️ Phase 1 — Pipeline Setup

✅ Groq API Key configured — LLM will be used for MITRE analysis

⏳ Click Load Pipeline to start...

📊 Phase 2 — Performance Evaluation

After loading the pipeline the status box above shows CV fold scores, mean accuracy, std, min and max — matching the notebook's Phase 2 output. Click below to render the confusion matrix (Phase 2.3).

Confusion Matrix — 5-fold Stratified CV

🧪 Phase 3 — Full Pipeline Test on a Sample

🔢 Dataset Index

Select a sample to analyze (0 – 15,035)

0 15035

3.1 — Prediction Summary

3.2 & 3.3 — SHAP & LIME Local Explanations

🔍 SHAP Local Explanation

SHAP — Feature Impact

⚡ LIME Local Explanation

LIME — Feature Impact

3.4 — DiCE Counterfactual Explanations

3.5 — MITRE ATT&CK Mapping + Security Report

📊 Model: RandomForestClassifier | 🔍 XAI: SHAP, LIME, DiCE-ML | 🎯 Threat Intel: MITRE ATT&CK | 🦙 LLM: Ollama (llama2) / Groq | 📦 Dataset: Drebin-215