Linear Support Vector Machine-Recursive Feature Elimination with Cross Validation untuk Menentukan Penciri Kerawanan Pangan di Jawa Barat

Abstract

Seleksi peubah dapat digunakan untuk mengetahui informasi mengenai penentuan tingkat kepentingan peubah yang menyebabkan suatu kejadian, salah satunya kerawanan pangan. Pengentasan kerawanan pangan merupakan tujuan kedua Sustainable Development Goals (SDGs), yaitu mengakhiri kelaparan yang diharapkan tercapai pada tahun 2030. Penelitian ini bertujuan mengukur kekuatan asosiasi peubah dengan kerawanan pangan serta mengidentifikasi peubah-peubah yang mencirikan kerawanan pangan di Jawa Barat menggunakan algoritma Support Vector Machine-Recursive Feature Elimination with Cross Validation (SVM-RFECV). Berdasarkan hasil penelitian, peubah yang memiliki asosiasi kuat dengan kerawanan pangan adalah tingkat pendidikan kepala rumah tangga (KRT) dan jumlah anggota rumah tangga (ART) yang memiliki tabungan. Umur KRT, status bekerja KRT, listrik, air minum layak, kepemilikan Jaminan Kesehatan Daerah (Jamkesda), akses Kredit Usaha Rakyat (KUR), dan jumlah ART memiliki asosiasi yang sangat lemah dengan kerawanan pangan. Peubah-peubah yang mencirikan kerawanan pangan di Jawa Barat adalah jumlah ART yang memiliki tabungan, KRT dengan pendidikan tertinggi SMP, jumlah buta huruf, jumlah ART, jumlah perokok, kepemilikan aset tanah, status bangunan dinas, penerima transfer, penerima Bantuan Pangan Non-Tunai (BPNT), kepemilikan BPJS Kesehatan, dan umur KRT. Banyaknya ART yang memiliki tabungan menempati urutan teratas peubah penciri kerawanan pangan di Jawa Barat.

Feature selection can be used to find out information about determining the level of importance of variables that cause an event, one of which is food insecurity. Alleviation of food insecurity is the second goal of the Sustainable Development Goals (SDGs), which aims to end world hunger by 2030. This study aims to measure the strength of the association of variables with food insecurity and identify the variables that characterize food insecurity in West Java using Support Vector Machine-Recursive Feature Elimination with Cross Validation (SVM-RFECV) algorithm. Based on the study’s results, the variables strongly associated with food insecurity are the education level of the head of the household and the number of household members who have savings. The age of the head of the household, the working status of the head of the household, electricity, decent drinking water, grantee of health insurance local program (Jamkesda), access to KUR, and the number of household members have very weak associations with food insecurity. The variables that characterize food insecurity in West Java are the number of household members who have savings, the head of the household with the highest education in junior high school, the number of illiterates, the number of household members, and the number of smokers. In addition, other characterized variables were ownership of land assets, ownership status of the building is an official house, transfer recipient, grantee of non cash social assistance (BPNT), grantee of health insurance national program (BPJS Kesehatan), and age of head of household. The number of household members who have savings ranks at the top of the variables that characterize food insecurity in West Java.