I have a series of land use maps at different times as a result of a classification process on Landsat images. I am mapping 3 different thematic classes : bare soils, vegetation and croplands.
I would like to answer the following question : "are land use change values significant ?" I believe chi-square test might help here.
How does the test work for my use case, and how should I run it ? Should i sample some pixels (if so, how much ?), or or work with the entire rasters ?