Evaluation and Analysis of Few Parametric and Nonparametric Classification Methods

Image processing is one of most popular research area. Image processing consists of a number of phases to be carried out, among which classification is one. Classification of an object is a vital job in computer vision area. Classification reflects the final results accuracies. So everyone has to pay attention at this phase. Lots of work is done on comparative learning of classification methods. This survey paper demonstrates study of parametric and non-parametric classification techniques. Study isolates parametric and nonparametric classification techniques which are employed in classification phase and offers tree depictions of such methods. For comparative study 9 classifiers are utilized among which 6 belongs to parametric category and remaining 3 belongs to non-parametric category.9 classifiers considered here are super simple, their interpretation is nice probabilistic, used mostly and well known. Assessment parameters considered are Kappa Statistic, Mean Absolute Error (MAE), Receiver Operating Characteristics Area (ROC Area) and Root Mean Square Error (RMSE). For validation test 10 cross fold method is considered. Results show that overall Decision Tree classifier or subtypes of decision tree classifier performs well. It gives better output to high values of Kappa Statistic and ROC Area measures. Also, it produces better results with less value of MAE and RMSE measures. The results produced by Bayesian Net, Naive Bayes techniques are similar.