Klasifikasi Malware Android Dengan Menggunakan Metode LightGBM Algoritma

By 2023, approximately 3.6 billion people worldwide were using Android smartphones. With a global market share of 71.75% by the end of 2022, Android stood out as the most widely used operating system. However, its success has made it a prime target for cybercrime, particularly through malicious actions such as malware attacks. Over time, malware has continued to evolve, becoming increasingly difficult to detect. Hence, the need for reliable detection methods arises. In the field of IT, machine learning has shown considerable efficiency in detecting malware. The author proposes the LightGBM Algorithm in Machine Learning as an approach to classify Android malware. Many boosting tools utilize pre-sorting-based algorithms (for instance, the native algorithm in XGBoost) in decision tree learning. While this is a simple solution, it's not easily optimized. On the other hand, LightGBM uses a Histogram-based algorithm, which discretizes continuous feature (attribute) values into bins, speeding up training and reducing memory usage. Therefore, the author proposes employing the LightGBM algorithm in this research. Upon conducting the research, the LightGBM Model achieved a Detection Accuracy of 96,37% and a Validation Accuracy of 96,34%. Its F1-Score stood at 0,9638.