Handling overdispersion in poisson regression using negative binomial regression for poverty case in west java
Felina Suryadi, Stanley Jonathan, Kelvin Jonatan, Margaretha Ohyver
Abstract
Poverty is one of government's problems in West Java Province that should be suppressed even abolished considering its violates human rights to live in prosperity. In September 2019, number of Poor People reaches 3.38 million people (6.82 percent). Thus, the results of this paper expected could help government in overcome poverty problem in West Java Province by finding the best regression method and model to predict number of poor people and find the most influential predictor that affects the number of poor people in West Java Province. We used secondary datas in 2019 from BPS Jawa Barat with four predictor: Province Minimum Wage (UMP), Human Development Index (IPM), Open Unemployment Rate (TPT), and Number of House Hold (RumahTangga) and one respon variable: Number of Poor People in West Java. The conclusion of this paper are: overdispersion could be overcome and modelled better using Negative Binomial, and the significant variables (which means have impact for response variable) in this case are only two predictor: Human Development Index (IPM) and Number of House Hold (RumahTangga). Human Development Index (IPM) is the most impactable predictor that government should build up to decrease Number of Poor People, while Number of House Hold (RumahTangga) does not have a really big impact but if it get depress would decrease Number of Poor People in West Java.