The stepwise logistic regression

The relationship between the indicator set and each land use were analysed by stepwise logistic regression, taking into consideration statistical problems as autocorrelation (by random sampling) and multi-collinearity (supervised variable elimination during the stepwise regression analyses). This led to an equation that was then used to construct land use change probability maps. The relationships found are a result of the area’s land use history. This is of course country specific, but these relations also change between areas with strong environmental, historical and cultural differences. For this reason, large nations were further split up in separately analysed zones on the basis of the farming system map published by FAO and Worldbank (2001).