فهرست نشریات علمی دسترسی آزاد ایران

In the realm of classification studies, existing literature indicates that, when the relationships among exploratory variables extend beyond linear functions, nonlinear classifiers tend to outperform their linear counterparts. This study employs concordance measures to attain optimal outcomes in a classification task.In this regard, we examine the connection copula among the exploratory variables, as well as the copula linking the exploratory attributes to the target attribute are taken into consideration. As a major novelty, our classification approach utilizes a convex combination of the pairwise Spearman’s rank correlation coefficient $\rho$ and the pairwise Kendall’s association $\tau$. Through a simulation analysis, we assess the performance of our algorithm, which demonstrates its superiority over alternatives, including copula-based classification methods as well as machine learning classification models. We also, provide an application of our method to the classification of COVID-19 dataset for more illustration.