A Performance Analysis of Predicting Diabetics Disease Using Hybrid Machine Learning Approach

Nowadays the diabetic, is considered a chronic disease. It is one of the metabolic disorders seeded by tenaciously accumulated blood sugar levels. If diabetes is early predicted then the risk is lowered. Due to the trivial quantity of tagged data or the lack of labels in the PIMA datasets, it is tremendously sticky to create a strong and infallible prediction of diabetes. In our work we proposed a XG Boost for diabetes prediction using k-cross validation, and compared with k-nearest neighbor, decision tree, Support vector machine, Ada Boost and naive bayes, and also unusual rejection, satisfying in absent values, data standardization, and characteristic selection is done. Using the Pima Indian Diabetes Dataset, each one of the investigations in this work was carried out under identical experimental settings. Our proposed ensembling classifier provides an experimental result of 85% compared to existing algorithms. Our proposed framework performs better than others in predicting diabetes.