A COMPARISON OF M-ESTIMATION AND S-ESTIMATION ON THE FACTORS AFFECTING IR DHF IN EAST JAVA IN 2017

Abstract

A COMPARISON OF M-ESTIMATION AND S-ESTIMATION ON THE FACTORS AFFECTING IR DHF IN EAST JAVA IN 2017

Mardiana 1 , Arief Wibowo 2 , Mahmudah 2 , Pipit Festi W 3

1 Bachelor Degree of Public Health Department, Faculty of Health and Pharmacy, Universitas Muhammadiyah of East Borneo, Borneo, Indonesia 2 Biostatistics and Demography Department, Master’s degree of Public Health Department, Faculty of Public Health, Universitas Airlangga, Surabaya, Indonesia 3 Faculty of Health Sciences, Universitas Muhammadiyah of Surabaya, Surabaya, Indonesia Correspondence Addres : Mardiana Email : mar348@umkt.ac.id, Phone : +6281351414752

ABSTRACT Robust regression on M estimation and S estimation is the Ordinary Least Square (OLS) regression on the data outlier. East Java is one of the provinces in Indonesia with a high case fatalitiy rate (1.34%). The raising of Dengue Haemoragic Fever (DHF) in East Java has been influenced by climate, population density, human behavior, and environmental sanitation. This study aimed to compare robust regression research by using M estimation and S estimation on the factors that affect IR DHF. This was done to get the best model regression on the data outlier based on the biggest R adjusted and the smallest MSE. This study utitlized observational research with a nonreactive research design using secondary data. The independent variable consisted of population density, healthy

behavior, healthy living environment house, and precipitation in East Java in 2017. The dependent variable was

incident

rate of DHF in 2017. The population included 38 regencies in East Java, while the sample was 35 regencies/cities selected using simple random sampling. The analysis used robust regression on M estimation and S estimation weighting by Tukey’s Bisquare. Robust regression on S estimation was found to be the best robust regression on data outlier with R 2 adjusted (0.996) and MSE (0.229). Robust regression on S estimation was 𝑦 ̂ = 54.826 + 0.011 (population density) – 0.136 (% healthy behavior) - 0,404 (% healthy house ) - 0,005 (precipitation). Some factors that affect IR DHF can be the main focus for the prevention and control of DHF for the government and society.

Keywords: robust regression, outlier, estimation, estimation, DHF