Reduction of Data Dimensions in The PLA Process

Abstract

Reducing dimensional data is an essential step before data analysis in Predictive Learning Analytics (PLA) for student dropouts. It was reducing dimensions in the study using the CATPCA method. CATPCA has advantages in reducing data dimensions on measurement variables of various levels such as nominal, ordinal, and numerical, which may not have a linear correlation between one variable and another, such as variables related to the PLA data processing. This study's results are five factors that store important information about the input variables, namely social and economic, academic program, institutional, academic performance, and personal. The conclusions of this study will be beneficial for further research in the PLA process