Applications Review of Hassanat Distance Metric
Ahmad B. Hassanat, Esra’a Alkafaween, Ahmad S. Tarawneh, Samir Elmougy
Abstract
Numerous machine learning methods depend on distance measures. In both supervised and unsupervised learning, these distance measures are primarily employed to assess the degree of similarity between data points. One of the most recent distance metrics that comes highly recommended by several academics for its improved performance in machine learning applications is the so-called Hassanat distance. In this paper, we review different applications that used this distance, improve its proof as being a distance metric, and simplify its formula to speed up the process. We found that this distance performs better when the data contains noise, outliers, and in particular different scale, and therefore, we recommend using it in some real-world machine learning applications.