PRINCIPAL COMPONENT ANALYSIS - AN EFFECTIVE FEATURE EXTRACTION METHOD IN HIGH DIMENSION SPACE

Abstract

When solving maths problems,  we often meet  sample sets  with  very high dimension space. Moreover,  in those sample sets there may be poor description parameters, mutually interactive  parameters that  makes problem resolution difficult. The matter is to extract features from description attributes in sample sets for the data mining process. This paper,  show that Principal Component Analysis (PCA) is an effective method for feature extraction in high dimension space. Experiments in forecasting the biological activity based on the sample set PDGFR (platelet derived growth factor receptor) Inhibitors have shown its effectiveness.