Improvement of Filter Model Feature Selection By Combining 3 Different Filter Model Feature Selection

27856

Nowadays, microarray data is an important contribution to the diagnosis of illness. High p and low n parameters that have micro‐array data, affect the machine learning algorithms classification success rate negatively. For this reason feature selection is a pre‐processing technique with great importance in microarray classification. In this study 3 different feature selection algorithms applied to 7 different datasets and results will be achieved by combining feature selection process. The resultsaretested with k‐fold cross validation by using k‐nearest neighbor (k‐NN) method. Classification success rate that obtained by the proposed method is compared withobtained by the filter modelled feature seleciton method individually. In addition, proccesing times are compared.

31

Micro‐arrray data set , Classificaiton , Feature Selection

Afyon Kocatepe University Journal Of Science an‎d Engineering

35

Afyon Kocatepe University Journal Of Science an‎d Engineering