๐ **Project Title:** Android Malware Classification Using Machine Learning
๐ **Dataset Used:** TUANDROMD (Tezpur University Android Malware Dataset)
๐ **Techniques Applied:** Static analysis, Random Forest, Logistic Regression
๐ **Models Compared:**
– Random Forest (Non-linear, 99% accuracy)
– Logistic Regression (Linear baseline, 98% accuracy)
๐ **Contents Covered in the Presentation:**
โข Research question and motivation
โข Dataset introduction (4,464 Android applications)
โข Feature types extracted from AndroidManifest.xml
โข Preprocessing and 60/40 train-test split
โข Model descriptions and evaluation metrics
โข Confusion matrix, ROC curve, feature importance
โข Key findings and what was learned
โข Limitations and future improvements
๐ฏ **Key Takeaway:**
Machine learning, particularly Random Forest, can accurately classify Android malware using static manifest-based features.
source
