Data Mining Classification Approach to Predict The Duration of Contraceptive Use

Abstract

Family planning program implementation in Indonesia has a plethora of challenges. One of the biggest challenges to implement the family planning program in Indonesia is the huge percentage of contraceptive discontinuation rates for around 29% in 2019. Based on that problem, the data mining classification approach is proposed to produce a model that can predict the duration of contraceptive use by productive couples. Through Cross-Industry Standard Process for Data Mining (CRISP-DM) process, it tested four experiments to seven data mining techniques with 39.594 contraceptives used histories dataset which is sourced from the Demography and Health Survey of Indonesia (DHS) in 2017. The result shows that the Adaboost data mining technique produced the best performance of contraceptive used prediction model, with the accuracy score of the classification model as 85.1%, precision score as 85.1%, recall score as 85.2%, and F1 as 85.1%. The model produced in this study can be used to estimate the length/duration of a particular type of contraceptive method which is used by each productive couple. That information is useful to prevent discontinuation potencies among contraceptive users for a further period.