Volume 18, No. 6, 2021

An Empirical Performance Evaluation And Comparison Of Different Classifiers On Standard Data Sets

Kamlesh Kumar Raghuvanshi , Subodh Kumar , Arun Agarwal


Supervised Machine Learning (SML) refers to the mapping of the input variable to output variable using an algorithm. The correctness of learning is the number of correct predictions/classifications after training a model. This study shows the behavior of different supervised classifiers namely distance-based non-parametric algorithm KNN, statistical-based Naïve Bayes classifier, parametric method SVM, and Neural Network on linear and non-linear data. The performance of the algorithms is evaluated for accuracy score parameter on some authentic dataset repository of UCI Machine Learning Repository and compared to find out which algorithm is best suitable for which type of data sets. One of the important steps is also included before analyzing the accuracy score. This is calculated using Standard scalar libraries present within Anaconda software.

Pages: 1995-2002

Keywords: Classification, K-Nearest Neighbor, Naïve Bayes, parametric, prediction, supervised, SVM

Full Text