Practical GIS

Fuzzifying crisp data

What we have now are three layers containing raw distance data. As these data are part of different criteria, we cannot directly compare them; we need to make them comparable first. We can do this by normalizing our data, which is also called fuzzification. Fuzzy values (μ) are unitless measures between 0 and 1, showing some kind of preference. In our case, they show suitability of the cells for a single criterion. As we discussed earlier, 0 means 0% (not suitable), while 1 means 100% (completely suitable).

The problem is that we need to model how values between the two edge cases compare to the normalized fuzzy values. For this, we can use a fuzzy membership function, which describes the relationship between raw data ...