CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUES

Android malware has grown steadily into a major internet threat. Despite efforts to identify and categorize malware in seemingly safe Android apps, addressing this issue is still lacking. Therefore, understanding the unique behaviors of common Android malware categories is essential. This study util...

Full description

Saved in:
Bibliographic Details
Main Authors: abdullah alsraratee, Ahmed Al-Azawei
Format: Article
Language:English
Published: Faculty of Engineering, University of Kufa 2025-04-01
Series:Mağallaẗ Al-kūfaẗ Al-handasiyyaẗ
Subjects:
Online Access:https://journal.uokufa.edu.iq/index.php/kje/article/view/16526
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850198629008015360
author abdullah alsraratee
Ahmed Al-Azawei
author_facet abdullah alsraratee
Ahmed Al-Azawei
author_sort abdullah alsraratee
collection DOAJ
description Android malware has grown steadily into a major internet threat. Despite efforts to identify and categorize malware in seemingly safe Android apps, addressing this issue is still lacking. Therefore, understanding the unique behaviors of common Android malware categories is essential. This study utilizes machine learning techniques namely, K-Nearest Neighbor, Random Forest and Decision Tree to classify Android malware based on dynamic analysis. As feature selection and reduction techniques, Mutual Information and Principle Component Analysis are used. The research analyzes a large dataset, containing fourteen primary malware categories using the CCCS-CIC-AndMal2020 dataset. Unlike previous research, the proposed method makes a balance between the number of features and classifiers’ performance, resulting in an overall detection accuracy of 98% in the fourteen analyzed categories and excluding 78.87% of the original dataset’s features. The research, thus, introduces an efficient Android malware detection method that reduces the computational cost and improves the classification accuracy.
format Article
id doaj-art-3c558de91a3843a8abc1cda4dea90e2b
institution OA Journals
issn 2071-5528
2523-0018
language English
publishDate 2025-04-01
publisher Faculty of Engineering, University of Kufa
record_format Article
series Mağallaẗ Al-kūfaẗ Al-handasiyyaẗ
spelling doaj-art-3c558de91a3843a8abc1cda4dea90e2b2025-08-20T02:12:49ZengFaculty of Engineering, University of KufaMağallaẗ Al-kūfaẗ Al-handasiyyaẗ2071-55282523-00182025-04-0116029611810.30572/2018/KJE/160206CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUESabdullah alsraratee0https://orcid.org/0009-0000-6565-5612Ahmed Al-Azawei1https://orcid.org/0000-0002-4121-2531College of Information Technology, University of Babylon, Babil, IraqCollege of Information Technology, University of Babylon, Babil, IraqAndroid malware has grown steadily into a major internet threat. Despite efforts to identify and categorize malware in seemingly safe Android apps, addressing this issue is still lacking. Therefore, understanding the unique behaviors of common Android malware categories is essential. This study utilizes machine learning techniques namely, K-Nearest Neighbor, Random Forest and Decision Tree to classify Android malware based on dynamic analysis. As feature selection and reduction techniques, Mutual Information and Principle Component Analysis are used. The research analyzes a large dataset, containing fourteen primary malware categories using the CCCS-CIC-AndMal2020 dataset. Unlike previous research, the proposed method makes a balance between the number of features and classifiers’ performance, resulting in an overall detection accuracy of 98% in the fourteen analyzed categories and excluding 78.87% of the original dataset’s features. The research, thus, introduces an efficient Android malware detection method that reduces the computational cost and improves the classification accuracy.https://journal.uokufa.edu.iq/index.php/kje/article/view/16526androidmalwaredynamic analysismachine learningmalware category classification
spellingShingle abdullah alsraratee
Ahmed Al-Azawei
CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUES
Mağallaẗ Al-kūfaẗ Al-handasiyyaẗ
android
malware
dynamic analysis
machine learning
malware category classification
title CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUES
title_full CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUES
title_fullStr CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUES
title_full_unstemmed CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUES
title_short CLASSIFYING ANDROID MALWARE CATEGORIES BASED ON DYNAMIC FEATURES: AN INTEGRATION OF FEATURE REDUCTION AND SELECTION TECHNIQUES
title_sort classifying android malware categories based on dynamic features an integration of feature reduction and selection techniques
topic android
malware
dynamic analysis
machine learning
malware category classification
url https://journal.uokufa.edu.iq/index.php/kje/article/view/16526
work_keys_str_mv AT abdullahalsraratee classifyingandroidmalwarecategoriesbasedondynamicfeaturesanintegrationoffeaturereductionandselectiontechniques
AT ahmedalazawei classifyingandroidmalwarecategoriesbasedondynamicfeaturesanintegrationoffeaturereductionandselectiontechniques