CART METHOD APPROACH AND HIGH DIMENSION SIMULATION DATA SELECTION AND RANDOM UNDER-SAMPLING METHOD IN STUNTING CASE

Stunting has become the main problem in every area for repair and investigation. Data modeling with To do classification could help for however many variables big is problematic in modeling classification and can complicate the interpretation process. Data modeling with an amount of enough significant variables could handle the selection process stepwise method. This study will create a classification model using tree decision Classification and Regression Tree (CART) with challenge amount variable predictor. A total of 26 variables and 650 observations were applied using the simulation data taken from real stunting data in Java east and the data used on the variable predictor which is stunting and normal categories. The result of the study is a method used that could propose selected variables in the stunting process with high accuracy. The acquired model has decisive information related to influencing factors stunting incidents by grouping family internal factors namely the health of parents who divides the knot root model decision as factor main in stunting incident.