An Ensemble Multinomial Naïve Bayes Classifier for Overlapping Prakriti Detection

The identification of ‘Prakriti’ types for the human body is a long-lost medical practice in finding the harmony between the nature of human beings and their physical, physiological, and psychological behavior. There are three fundamental ‘Prakriti’ types, i.e., ‘VATT-Dosha’, ‘PITT-Dosha’, and ‘KAPH-Dosha’. Most of the earlier machine learning techniques, like KNN, SVM, etc., were able to successfully classify an individual as one of the ‘Prakriti’ types. However, an individual may possess more than one ’dosha’, resulting in overlapping ‘Prakriti’ types. This research aims to address the challenging problem of overlapping ‘Prakriti’ clusters. First, we cluster the data into seven overlapping clusters using the k-modes method. Using this clustered data, we developed an ensemble classifier, which then classifies any given instance into the appropriate ‘Prakriti’ class. The performance evaluation is demonstrated by the multinomial naïve bayes classifier and the ensemble of multinomial naïve bayes classifier.