Determination of Important Variables in Divorce Type Classification Using the Random Forest Method with SMOTE Dania Siregar, Bintang Mahesa Wardana, Ahmad Syauqi Baihaqy, Liswatun Naimah, Dini Safitri, Qorry Meidianingsih, Almira Nindya Putri
Universitas Negeri Jakarta
Abstract
Central Jakarta are a very strategically located region in the heart of the capital city of Indonesia and is the center of the city, government, history, tourism, and elite malls with easy access to various buffer areas of the city. This variety of facilities, however, does not guarantee perpetuity in domestic life in the region. This fact can easily be seen from the steadily increasing divorce rate in the region since 2017. An interesting thing to notice is that there is noticeably more divorce lawsuit filed by the wife than the husband. There are various factors that may lead to a divorce lawsuit such as continuous disputes and quarrels, economic factors, and domestic violence. And those factors certainly cannot be separated from the individual background of the married couple such as age, profession, the level of education, and the length of marriage. The purpose of this study is to determine the level of importance of the variables used to classify the type of divorce lawsuit filed in Central Jakarta using Random Forest. Random Forest is a development of the CART (Classification and Regression Tree) method obtained by applying bootstrap aggregating and random feature selection to a standard CART method. The results of this study indicate that the most important variables that may affect the type of divorce lawsuit filed is the age and profession of both the plaintiff and the defendant followed by the age of the marriage. The accuracy of the classification of lawsuit filed by the wife and the husband reaches around 90% and 70% respectively.
Keywords: Clasification, Random Forest, SMOTE, Divorce
