Optimasi Hyperparameter Random Forest Classifier Menggunakan GridSearchCV Pada Sistem Prediksi Klasifikasi Aplikasi Malware Android

The growing use of smartphones, particularly those running Android, has increased public awareness of the security risks posed by viruses and malware. In order to achieve perfect malware detection and minimize the gaps in machine learning-based classification, approaches for accurate identification and classification still require refinement, even if machine learning models have been built for malware prediction in this field. Prior research has shown detection accuracies ranging from 93% to 95%, suggesting potential for improvement. This work suggests enhancing the Random Forest method by adding a grid search algorithm, which wasn't used in earlier research, in order to maximize hyperparameters. This study's major goal is to significantly increase categorization accuracy. We successfully showed that the suggested technique is superior by detecting malware-infected apps with an almost flawless 99% accuracy rate. With a near-perfect identification rate, the suggested approach represents a significant advancement over current models, which were only able to achieve a maximum accuracy rate of 95% on the same dataset. This study's method achieves excellent accuracy and offers a fresh way around the platform's drawbacks, particularly in situations where program processing resources are constrained. This research validates the efficacy of the enhanced Random Forest algorithm, signifying a paradigm shift in the identification of malware and enhanced cybersecurity protocols for the swiftly expanding smartphone industry.