Poverty mapping spatial regression analysis of settlement data from Somalia

Abstract

The study briefly reviewed spatial regression models for describing and explaining spatial data variation. One of the regression models, the spatial Durbin model (SDM) was used to identify possible causes of poverty in Somalia using somalia settlement census data. The exploratory spatial data analysis (ESDA) was used to identify data properties. The output from ESDA provide input into spatial modelling. Modelling involved specification, parameter estimation and hypotheses testing, computation of confidence intervals and assessment of goodness of fit. The Global and Local Moran's I were computed and it which showed dissimilarity of neighbouring settlements in North West Somalia and similarity of neighbouring settlements in North East and South Central Somalia with respect to the variable of interest, the proportion of families who cannot afford two meals per day, which was taken as a proxy indicator for poverty level. Finally implication of the findings on policy decision making for development planning was discussed and possible areas of further research recommended.