PROGNOSTICATION OF DIABETES USING DATA MINING MODELS

Background: Diabetes is an assimilative disease that could cause soaring in the levels of blood sugar. This un diagnosed rise in blood sugar from diabetes could damage a human beings nerves, eyes, kidneys, and other internal organs. Objectives: The following study is conducted in order to pinpoint and bestow a method for prompt prognostication of diabetes by making use of data mining tools. Methods: The data used was gathered under the Countenance of a doctor by using resolute questionnaires among patients of Sylhet Diabetes Hospital which is situated in Sylhet, Bangladesh. 16 instances were accumulated in this dataset. Algorithms namely Bagging, Sequential Minimal Optimization (SMO), Naïve Bayes, Instance Based method based on K-neighbor's (IBK), Random forest are used for comparison of the prediction of diabetes. Result: On collation of the 5 algorithms used WEKA, which is a data mining tool, the accuracy has been calculated for each of the algorithms with respect to their Tp-rate, Fp-rate, precision and recall rates. Conclusion: On appraisal of the results procured, it provides the ability to come to a culmination that the IBK algorithm is the best algorithm based on the accuracy and performance, while comparing it with the other models.