Comparison of Regression Methods in Permission Based Android Malware Detection

Şahin D. Ö., Kural O. E., Akleylek S., Kılıç E.

28th Signal Processing and Communications Applications Conference (SIU), ELECTR NETWORK, 5 - 07 October 2020 identifier

  • Publication Type: Conference Paper / Full Text
  • Keywords: Android malware, mobile malware analysis, static analysis, regression analysis, machine learning
  • Ondokuz Mayıs University Affiliated: Yes


In this study, applications developed for Android platforms are tested by static analysis based on machine learning. Permissions that have an important place in the security of the Android operating system are used as attributes. Using the regression techniques, which are among the types of machine learning, the applications are tested. Four different regression techniques are used in this study. These are linear regression, multilayered neural network, additive regression and regression techniques based on sequential minimal optimization. As a result of 10 cross-validations, the best result is obtained by linear regression, while the worst result is obtained by regression techniques based on sequential minimal optimization. The result obtained from linear regression is 0.8655 according to the Pearson correlation coefficient.