Automatically identifying suitable rulebase parameters in the context of solving the map overlay problem

Abstract

Analysis of geographically related data often requires the combination of data from different sources. Data are commonly represented in grids, and unfortunately, the grids containing different data do not match properly: they can differ in cell size and/or orientation. A novel methodology was presented to allow the data of one grid to be remapped onto the other grid. The method makes use of a fuzzy inference system that performs the remapping, using additional information relating to the data distribution. Previous research has revealed that the best parameters used in the inference system are dependent on the input, and as such an automatic determination of which parameters should be used, would improve the performance. In this article, we propose a solution for this automatic detection, by first generating a training set that is related to the input and then determining what the best parameters are for this training set.