Adolescents Psychological Well-Being Estimation Based on a Data Mining Algorithm

Abstract

Control of the risks for reducing mental health and psychological well-being of young people allows making timely managerial decisions aimed at reducing social tensions and increasing the safety of communities. Effective implementation of projects at the national and regional level is possible if there is relevant and dynamically updated information on the state of mental health of young people. The authors develop a special questionnaire for gathering initial data on psychological wellbeing of adolescents. However, for final conclusion about wellbeing, a qualified psychologist is needed who is not always available for organizations (especially for rural schools). In this regard, the use of methods of machine learning and data mining to create software that automatically assesses well-being according to results of respondents' responses is relevant. Within this study, the group method of data handling (GMDH) is used. The algorithm of twice-multilayered modified polynomial neural network with active neurons is applied to construct classifiers for 4 classes of well-being of schoolchildren. The data contain responses of about 200 adolescents aged 12–17 years from 11 rural schools. The results of this study demonstrate the percentage of correct classification for the two extreme classes of well-being (“well-being”, “not well-being”) not worse than 90% for an independent control sample of data.