Comparative Study of KDD and CRISP-DM Methodologies for Phishing Identification

Numerous studies have been conducted due to the high value of user data, which has led to fake (or cloned) websites known as “phishing” websites. In this context, this work aimed to experimentally compare the KDD and CRISP-DM methodologies used for phishing identification in a publicly available dataset. Experimental comparisons of machine learning algorithms and approaches were carried out, with competitive outcomes. When using the Perceptron Multilayer technique to create CRISP-DM, we received a 92.9% F1 score. Additionally, a comparison of the phases between the two approaches revealed some insightful qualitative findings.