Effective Use of Naïve Bayes, Decision Tree, and Random Forest Techniques for Analysis of Chronic Kidney Disease

Abstract

The researcher is using a classification method for chronic kidney patient analysis of data. Chronic kidney disease data contains 25 attributes and 400 instances. Now, we proposed a best model by applying the decision support system of naive Bayes, decision tree J48 algorithm, and random forest classifier techniques, and really, this proposed model will be helpful to predict the further CKD as well as not CKD patients on the basis of different parameters. During analysis, the naive Bayes classifier correctly classifies instances with the 97.50% accuracy, decision tree J48 algorithm finds the correctly classified instances with the 98.33% accuracy, and similarly, random forest is analyzed and giving output with 100% accuracy and 0% of incorrectly classified instances. Therefore, the random forest decision tree classifier algorithm is the best and produces the most accurate and correct results with the 70 percent of the split value (train on a portion of the data and test on the remainder), and the value of ROC area is 1%. The main objective of the research paper is comparative study of NB classifier, DT J48, and RF to analyze chronic kidney disease (CKD) patient’s data and to predict how many patients are having CKD. An analysis of how many patients currently have kidney disease and how many people may have this disease in the future has been attributed to it. When analyzing the same algorithm, the decision tree J48 shows that the tree variant can be 100% diagnosed with kidney disease in the future, and the random forest algorithm has analyzed it 100%.