RENTAKA: Detecting Ransomware at Pre-Attack Stage Using Machine Learning Approach

In this work, we propose the analysis and evaluation of various machine learning classifiers over RENTAKA dataset to detect crypto-ransomware in its pre-encryption phase. A full pre-encryption activity dataset was used to evaluate the five classifiers: Random Forest, Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), Logistic Regression and Decision Tree. The Random Forest Classifier outperformed with 96.29% accuracy which is closely followed by SVM and Logistic Regression of 94.98%. By comparing the performance with the PEDA baseline, the framework clearly showed its advantage by not severely increasing the False Positive Rate (FPR), which proved the RENTAKA efficiency in avoiding false alarms. For performance evaluation, in addition to accuracy, metrics like Precision, Recall, F1-Score, and AUC-ROC were also considered to analyze detection capabilities in a balanced way. The outcomes indicate RENTAKA's robustness and its promise as a proactive defense mechanism against ransomware. The paper ends with a conclusion about the impact of early diagnosis, and future improvements.