EVALUATING THE EFFECTIVENESS OF PREDICTION TECHNIQUES FOR CYBERATTACKS: A COMPREHESIVE TAXONOMY

The rising threat of cyberattacks in today's society emphasizes the urgent need for improved methods to both detect and prevent these incidents. This paper focuses on assessing the effectiveness of various techniques for predicting cyberattacks. The DTCF taxonomy was proposed for predicting these attacks, considering datasets, techniques, challenges, and future trends. This taxonomy includes four key stages. 1) data preprocessing, 2) feature selection, 3) development of prediction models, and 4) their subsequent validation and assessment. Our research reviews progress algorithms for each stage, analyzing their advantages and weaknesses. Consequently, the results of this study emphasize the critical role of precise detection and prediction in combating the increasingly complex threat of multiple cyberattacks, which are inherently more challenging to identify and predict than isolated incidents. Our examination of diverse learning methods reveals the essential role of data preprocessing in enhancing the efficacy of prediction systems. Effective preprocessing aids in reducing issues like noise, outliers, missing data, and extraneous features and, by doing so, refining the accuracy of predictions.