ANALYZING THE PREDICTIVE PERFORMANCE OF CLASSIFICATION ALGORITHMS USING SUPERVISED LEARNING MODELS BASED ON SPLITTING RATIO

In India more than 30 million people have now been diagnoses with diabetes. The Crude prevalence rate is thought to be 9 percent in India's urban areas. The incidence in rural areas is about 3 percent. In this paper we address the objective is likely to develop a model for diabetes with>70% accuracy using salient classification algorithms in the evolution of the learning models. We employing with the problem comprises of 768 observations of medical details for Pima Indian patient records. The major concern is each record has a class value that indicates whether the patient suffered on onset of diabetes within stipulated time is the measurement. Consequently this research recommends categorizing the whole process into data handling, data summarizing and making predictions. Finally we showed that accuracy of the predictions made for a test dataset as the percentage correct out of all notable predictions made on the dataset with splitting ratio and hence determine the performance among the Naïve Bayes ,Random forests and Logistic regression with balanced classes