Comparison of Kneighborsclassifier with RF Algorithm to Improve Accuracy for Phishing Attack Detection in Website

The objective of this study is to improve the detection of phishing attacks, which aim to steal sensitive information from unsuspecting users, by comparing two machine learning algorithms: k-Nearest Neighbors (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{k - N N}$</tex>) and RF. In the study, we used a dataset of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 2 5}$</tex> images to evaluate both algorithms. Out of these images, 70% (or 130 frames) were allocated for training the models, while the remaining 30% were used for testing their performance. Each algorithm was tested through 20 separate iterations to ensure reliability. The accuracy of each algorithm was assessed using statistical analysis, with a G-Power value set at <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{8 0 \%}$</tex> for significance. The RF algorithm showed superior performance, achieving a high detection accuracy rate of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 0. 5 4 \%}$</tex> for phishing attacks. This result was statistically significant with a p-value of 0.04. In comparison, the k-NN algorithm did not perform as well. Overall, the study finds that the RF algorithm is more effective than k-NN in detecting phishing attacks, making it a better choice for enhancing cybersecurity measures.