|
Poverty is one of social topic which will be talked and disputed continously at national and international forum. The fact shows that development which has done can not reduce indigent population in the world. To reduce poverty, we have to know who ingoing in indigent category and where they stay behind. MARS is a classification method which are innovative and relative flexible to investigate relationship pattern among dependent and independent variables without any assumption to its functional form.This research investigates the model accuration of MARS for village or sub-district's status, indigent (code 1) or not indigent (code 0), obtained by PSE'05.There are 27 predictor variables obtained by PODES SE'06. This research shows that the accuration of classification in indigent sub-district for urban region is 100 percents, affected by 17 community factors, i.e., distance to nearby post office, ratio of families live in flood gristle, healthy card receiver family percentage, ratio of health facility per 1000 citizen, total industrial enterprise, distance to nearby market, distance of village to the capital regency, agricultural family percentage, pra-KS and sejahtera 1 family percentage, distance of village to other capital regency, ratio of non PLN electricity user, total market, ratio of families, live in dirty residence, family having indigent card percentage, unskilled laborer in agriculture percentage, total telephone-shop, and distance of village to the capital district. |
