ANALYSIS OF FACTOR BASED DATA MINING TECHNIQUES

Factor analysis, which is a regression based data mining technique, used to represent a set of observed variablesin terms of common factors. This paper explores the key properties of three factor based techniques viz. principal componentregression, generalized least square regression, and maximum likelihood method and study their predictive performance ontheoretical as well as on experimental basis. The issues such as variance of estimators, normality of distributed variance ofresiduals, effect of multicollinearity, error of specification, and error of measurement are addressed while comparing theirpredictive ability.