Data Mining Probabilistic Classifiers for Extracting Knowledge from Maternal Health Datasets

Abstract

Data Mining is an important sub-process of Knowledge Discovery in Databases (KDD) or Knowledge Discovery Process (KDP) methodology that is mainly used for applying various data mining techniques and algorithms on the target data. In this research paper, the authors have made an attempt to discover knowledge by classifying the maternal healthcare data of Jammu and Kashmir State of India (now declared as Union Territory by the Government of India). The data for the present research work was collected from a web portal named as Health Management Information System (HMIS) facilitated by Ministry of Health and Family Welfare (MoHFW), Government of India. The data consists of diverse health parameters pertaining to the maternal health of women and for this study, the maternal healthcare data of all districts of Jammu and Kashmir State was considered. Two data mining classifiers viz. Bayesian TAN and Naïve Bayes were applied for classifying the districts of Jammu and Kashmir State into High MMR and Low MMR districts based on the available past data from 2014 to 2018. Additionally, evaluation measures viz. Accuracy, F-measure, Area under the Curve (AUC), and Gini have been used to evaluate the performance of the models developed by Bayesian TAN and Naïve Bayes.