Determining minimum feature number of classification on clear cell renal cell carcinoma clinical dataset

Kidney plays an important role in human bodies. It maintains homeostasis and removes some harmful substance by making and ejecting urine. Renal cell carcinoma, especially clear cell renal cell carcinoma (ccRCC), is the most common type of kidney disease that accounts for 2~3% of human malignancies. Early diagnosis and accurate classification of ccRCC is an important factor to decrease the motility rate. In this paper, the classification methods combined with feature selection have been examined on ccRCC clinical dataset. The data were obtained from The Cancer Genome Atlas (TCGA) that offers various cancer datasets including clinical data. We implemented 7 feature selection methods to identify important clinical factors and applied those factors into 4 classification algorithms such as C4.5, support vector machines, artificial neural network and Bayesian network. The experimental results showed that the Bayesian network with top 15 features was the most suitable prediction model. Moreover, the sensitivity was increased by employing feature selection approaches for classification of ccRCC data.