Perbandingan CART dan SMOTE CART dalam Mengklasifikasikan Kebutuhan KB Tidak Terpenuhi di Indonesia

Abstract

Indonesia is ranked fourth in the world as the country with the largest population. The high population growth in Indonesia can cause problems in several fields. The government seeks to suppress the rate of population growth through the Family Planning (KB) program. In Indonesia, the number of unmet needs for family planning is still relatively high and has not yet reached the BKKBN target. Therefore, it is necessary to identify the characteristics of unmet need for family planning among married women or living with partner. This study used the Classification and Regression Trees (CART) method. This study handling unbalanced data by Synthetic Minority Oversampling Technique (SMOTE). This study aims to compare the performance of the CART and SMOTE CART classification methods in classifying unmet need for family planning and to identify the characteristics of unmet need for family planning among married women or living with partner in Indonesia. The SMOTE CART model has better performance than the CART model, with the percentages of balanced accuracy, sensitivity, and specificity being respectively 54.83%, 34.96%, and 74.70%. In general, the characteristics of unmet need for family planning among married women or living with partner in Indonesia are having 1-4 living children, not getting information from mass media, not accessing the internet in the last month, having a primary or secondary education level, a husband with no education or with a primary or secondary education level, and aged more than 30 years old. Keywords: CART, SMOTE CART, unmet need for family planning