Improved K-Means with Adaptive Divergence Weight Bat Algorithm (IKM-ADWBA) and Feature Selection for Type 2 Diabetes Mellitus Prediction

Abstract

Increase in blood glucose (hyperglycaemia) leads to Diabetes Mellitus. There are two kinds of Diabetes mellitus: (Type 1 Diabetes Mellitus (T1DM) and (Diabetes Mellitus (T2DM), then former one is dependent on insulin and the latter one is independent of insulin. Various factors make it difficult to diagnose it. SO the author focuses at binging-in and analyzing the method for making a novel robust diagnosis system using data mining methods. Complete datasets is necessary for data mining techniques, but these techniques doesn’t give accurate results with missing values and all features. So, for prediction, Handling Missing value replacement and selection of important features are becomes a major issue. Hence, Adaptive Neuro Fuzzy Inference System (ANFIS) were proposed to acquire the missing value in dataset and to rectify the above mentioned issue. Then for an effective seed selection in Improved K-means algorithm, Enhanced Inertia Weight Binary Bat Algorithm (EIWBBA) is proposed, which results in high convergence speed. This research work proposed for feature selection with the help of Improved Distributed Kernel based Principal Component analysis (IDKPCA) with less time, after minimizing the entire feature space to the best features set. Then for classification of clustered samples, the author brought-in the Support Vector Machine (SVM). The experimental result confirms that the proposed algorithm gives the best classification accuracy rate when compared with other methods. From Pima Indians Diabetes, the data set has been considered and the experiment is done with the help of MATLAB for examining the Knowledge and the results were distinguished with other outcomes using appropriate toolkits.