Prediction Of Type-Ii Diabetes Using Machine Learning Techniques

Abstract

In the world, 70% of the death rate is due to the non-communicable Diabetes disease. Due to unhealthy lifestyle, the majority 90 to 95% diabetes cases are Type 2 diabetes which can be determined by the examination of diabetes-related parameters. The main propose of this study to developing a fuzzy expert technique to diagnosis of diabetes mellitus very efficiently. The implementation of this techniques is involved four main steps like (a) Fuzzification (b) Rules Evaluation (c) Output aggregation and (d)Defuzzification. The two comparative studies are done in this work. First, the proposed techniques are compared with regression method and several classification techniques such as Native Bayes, Support Vector Machine and Multilayer Perception. Second, use Mamdani fuzzy interference method to diagnosis the type II diabetics mellitus effectively. The fuzzy expert techniques were developed and validated with original data using data mining algorithm. The Pima Indian dataset, includes 768 records and 9 attributes were used in this research work. The pre-processing method are used two different techniques (1) Multiple Imputation method (2) Listwise deletion method to handled missing data in the given dataset. Finally, the different evaluation metrices are calculated and compared with results.