Proposed model to balance between accuracy and efficiency in the detection of phishing: An approach that combines clustering and random forest

Abstract

The purpose of this study isto improve the identificationof phishing attempts by using a complete method thatintegrates clustering pre-processing with efficient RandomForest training techniques.Applying clustering algorithmsto a well selected phishingdataset enables the identification of patterns and the refinement of features, resultingin enhanced accuracy in detecting phishing attempts. Theresearch concurrently investigates methods to decrease thetraining duration of RandomForest, such as modifying thequantity of weak learners, using sampling approaches, andexamining different fusing procedures. The study endeavorsto achieve a harmonious equilibrium between precision andeffectiveness via a process ofrepeated refinement. The results enhance the area of cybersecurity by providing valuable insights into the effectiveuse of clustering and RandomForest training to mitigatephishing threats with greaterresilience.