An Experimental Study of Diversity of Diabetes Disease Features by Bagging and Boosting Ensemble Method with Rule Based Machine Learning Classifier Algorithms

Energy produces by blood sugar in human body. Diabetes mellitus is a condition in which human body cannot manage energy. Blood glucose, Insulin etc., and their functions become unmanaged in the glucose energy system in blood. This unbalanced system generates many dangerous diseases as like blood pressure, diabetes etc., in body. Many different recourses of energy are available in the nature. In this paper we present dataset with their pattern by box whisker plot, histograms and extract nine best features by Chi-Square and plot correlation matrix for correlation of each feature with their heat_map. We have applied rule-based classification algorithms as Decision table, OneR and JRIP on prepared dataset. In proposed work we have managed these three selected algorithms by Bagging and Boosting ensemble methods. These ensemble methods calculate Classification Accuracy, Precision, Recall and F1-Score for diabetes UCI dataset. The summary with the results, finally we concluded that the highest accuracy (98%) of Bagging Ensemble Method and the accuracy, precision, recall and f1-score of the Decision Table, OneR, Jrip and Boosting algorithms were carrying less.