Doing ANN land suitability, input year land raster data and a variety of driving factors (precipitation, slope...), and the output layer for the suitability of probability of various land types, don't understand, the output of the output layer data is checked in and what data, modify the link weight gain error, strives for the great god give advice or comments