how do i know ,rules automatically created for FIS are good enough and crispy , there is graphical representation of rules in the form of curves

I have created fuzzy inference system for classification using sugeno type FIS. as i gave training set it automatically created membership functions by itself, how do i know these functions are good enough for classification. is writing rules is compulsory? or without writing rules i can proceed for testing process -shantala

Answer 1

[Arkadiy Turevskiy]

have you read through this example in the doc?

It explains that when you train your system, the tool automatically generates the model structure (including rules) for you. You can view the generated rules.

To see how good the created model is, try it on a validation data set and see if it produces reasonable results.