Penerapan Metode CART dan Agregat Bootstrap dalam Klasifikasi Status Gizi Balita Di Kecamatan Taman Sari, Kabupaten Bogor.

Abstract

Nutrition problem is a multidimensional problem that affect health and productivity of Indonesian society. The number of short nutritional status (stunting) and high malnutrition is a serious health problem because it picturizes life for the next 100 years. About 90% of the short children problems in the world occur in 36 countries, and the rank of Indonesia is number five. The purpose of this study is to determine the factors that affect the nutritional status of children and evaluate the stability of classification trees with bootstrap aggregating method.The research is conducted at Taman Sari subdistrict, Bogor with the number of samples 120 mothers who have toddlers. Statistical analysis used to analyze the factors that influence the nutritional status of children is the method of CART (Classification and Regression Tree). CART is a method of data exploration which uses decision tree technique. This study also applies bagging method (bootstrap aggrergating) to evaluate the stability of classification trees generated by CART. The results of the classification tree formed produces optimum tree with eight terminal nodes. Explanatory variables into the main node is the nutritional behavior. Level of accuracy for the CART method was 69.17%.The percentage of the classification on the delimitation of the best z using z-score -2.5 SD with level of accuracy 76.67% and 88.33%, and z-score -3 SDwith level of accuracy 73.33% and 79.17%. Bagging method has been able to evaluate the stability of classification trees by 79.28%.