Diabetes diagnosis rule-base reduction based on Topsis method

Diabetes is a group of metabolic disease caused by the imbalance between secreted and needed insulin levels. In the present study, all probabilities were examined considering eight (8) indications of diabetes and these indications binary values construct a rule base. Thanks to Clementine, support, confidence and lift for each rule are measurable, and then by Entropy method the criteria were weighted. Rules were ranked by Topsis method, so there is no need to check all the rules. At the end the rules are tested on test data and the results are given.

Keywords:

diabete , TOPSIS , association rule , expert system