In this tool our main aim being, static analysis of the apk, we have created the tool such that each apk is picked up along with all the embedded apks and the permissions are extracted. This extracted information is saved to a csv which is fed to the trained ML model that will predict whether the application is malicious or benign. Along with this the results from virus total is also shown.
Around 15k android applications, along with all the embedded apks are used for collecting the datasets. The permissions accessed by all these android applications is extracted and stored to a csv file in the form of onehot encoding. This dataset is used as the feature to train Support vector machines(SVM) with help of Sequential minimal optimization(SMO) method.
- Extraction of permissions from the apks.
- Work on training the model for the phases.
- Apktool
- Androlyze
- pickle
- pandas
- numpy
- sklearn
Python3 <script.py>