Data Mining
H. Koosha; S. Dangkoub; A. Barzanooni
Abstract
Background and Objective: Student mental health data has been recorded in the information systems of the universities across the country for several years, and due to its high volume, conventional statistical and psychoanalytic methods to predict patterns and factors affecting students' mental health ...
Read More
Background and Objective: Student mental health data has been recorded in the information systems of the universities across the country for several years, and due to its high volume, conventional statistical and psychoanalytic methods to predict patterns and factors affecting students' mental health are not effective. This is where data mining technology comes in handy and helps to predict and identify those at high risk based on the recorded data set of students 'physical and especially mental health status, and to make appropriate and timely decisions to improve students' condition. One of the main objectives of every managers of educational centers is making improvements in students’ educational performance. Besides the educational factors, physical and mental health is considerable which has a significant effect on students’ behavior. Therefore, some rules and patterns are required to make the best decisions, based on the prediction of students’ mental health state. This paper proposes a data mining approach for analyzing and extracting patterns in terms of new students’ mental health, which means whether they need to visit a psychologist. Our effort was on extracting hidden rules in new students’ mental health examination by employing classification approach. Methods:Techniques used in this study are decision tree, rule based classifier, neural network, logistic regression and support vector machine. Moreover, a parameter tuning process is done for all the techniques mentioned and the results presents the list of symptoms of individuals who need detailed examination. Findings:The results of the research represent that one can predict the status of students’ mental helath based on propsed model. One of the outcomes of decision tree is that if a person severely feels disappointed or seems to be obsessive by others, or feels that life is worthless, definitely a consultaion is needed. Conclusion: Considering that most of the existing research in the field of health data mining have focused on physical health, it is suggested that for future studies, all levels of health, i.e dimensions of students' health, including physical, social and spiritual health, as well as a combination of these dimensions be considered. In addition, a review of the various approaches and techniques appropriate to the psychological data set should be conducted with the aim of creating an appropriate classification for the existing techniques in this field. It is also suggested that the present data set or similar data sets (student health monitoring information) be examined with other classification techniques and the results be compared with the results of the present study. In general, it is suggested that data mining technology be used to extract hidden patterns in the mental health data set of school students at different levels of education, office workers and organizations. Finally, it is recommended that future research in this field first implement the clustering approach on the psychological data set and then use the classification and forecasting approaches.
