Predicting the performance of the players in NBA Players by divided regression analysis

Abstract

A divided regression model is built to predict the performance of the players in the National Basketball Association (NBA) from year 1997 until year 2017. The whole data set is divided into five groups of sub data sets and multiple linear regression model is employed to model each of the sub data set. In addition, the relationships among independent variables are checked by using variance inflation factor (VIF) to identify the risk of having multicollinearity in the data. Moreover, non-linearity of regression model, non-constancy of error variance and non-normality of error terms are investigated by plotting residual plots and quantile-quantile plots. Finally, a divided regression model is built by combining the results obtained from the sub data sets and the performance of the divided regression model is verified.