Predictive analysis using classification techniques in healthcare domain

Abstract

The main objective behind data mining applications is to specify that data, a fact, number, text etc. can be processed by a software system which results out as a useful knowledge. Data mining is interactive and iterative process. It is a discovery of association changes, automatic and semi-automatic patterns, anomalies, different structures and also events in data. The main purpose behind the implementation of data mining classification techniques on mental health care data set is to develop an automated tool for recognition, identification and publication of relevant mental health care information. This paper aims to help experts in healthcare domain in decision making by doing predictive analysis on mental healthcare dataset using classifiers in weka. We have mainly applied 3 classifiers- Naïve Bayes, J48 and Multilayer Perceptron. Naïve Bayes is an advanced form of Bayesian’s theorem, J48 is a decision tree based approach and Multilayer Perceptron is the simplest form in Neural networks. Dataset to be supplied to weka is Mental Healthcare survey with respect to IT industry all around the world. Data mining thus improves the quality of decision making process in its various applicative domains. Finally, this paper concludes by determining the major objective by illustrating data mining techniques and processes, methodologies and also the performance and accuracy observed in determining the best possible result from each existing technique so as to get the authentic information from the data set that we have supplied.