I am working on the Brazilian censuses 1980, 1991, 2000 and 2010.
I would like to merge some “small” adjancent municipalities with few observations.
I wanted to merge some of them belonging to a common mesorregion since, in theory, a mesorregion is a group of municipalities. Unfortunately, it appears that many muncipalities belong to several mesorregions at a time in the data.
Otherwise, I could perhaps detect municipalities with a common border by spotting groups of individuals indentified by the variable STRATA that would belong to several muncipalities. However, I am not sure this is a valid procedure.
Do you have any suggestions?
Thanks for your help.