Skip to Main Content
It is worthwhile to point out the fact that nature of given data plays considerable role in classifying the data accurately. To select an appropriate classifier for certain type of data, we are required to understand the behavior of classifiers on different data characteristics. The varying dimensions, number of instances, class labels, data correlation, and data distribution on different data classes, might characterize the data. In this study, the performance and behavior of five different supervised machine learning classification techniques have been investigated using six real life datasets that are taken form UCI Machine Learning repository along with artificially generated data. In the end, we have come up with some conclusions and findings which will be very supportive for upcoming researchers to develop a better understanding about data characteristics in combination with classifier's performance.