A Brief Review Of Geoinformatics Analysis On Poverty Data In Indonesia

Abstract

A research group on geoinformatics was built in 2010 at Department of Statistics of Bogor Agricultural University, Indonesia, to implement statistical analysis on spatial data. The initial step was to compile some statistical methods related on geographical regression of the simple approach and the complex ones. The methods were implemented to analyse the poverty data in Indonesia. Outcome of selected models was on poverty indicators in a district, such as percentage, expenditure per capita. The outcome was a priory influenced by some regional factors, i.e. local government policy, agro-climate typology, as well as local socio-culture. Therefore, the type of data may produce problems of outliers, outcome dependency, and non-stationarity. Classical approaches assuming there is no effect of the regional differences are not valid any more. For the first phase, the group then implemented methods related on GWR (Geographically Weighted Regression) including simultaneously or conditionally autoregressive models. To obtain firm statistical conclusion, the selected models were contrasted to the ordinary regression. The result indicated that district or regions affected on the poverty level. Hence, spatial factors cannot be neglected in analysing poverty in Indonesia. Additionally, geographical regression performed better than the ordinary models.