Prediction of Mental Health Among Iranian Adolescents Based on Nutrition, Physical Activity and Leisure Time Using Data Mining Techniques

The present research was conducted to predict mental health based on three factors: nutrition, activity, and leisure time, among students in the adolescent age group, using data mining techniques.

The present analytical study was conducted on 14274 data available in the Caspian 5 database. According to the CRISP-DM method, data mining was done in 6 steps using decision trees, k nearest neighbors, simple Bayesian and random forest techniques in Rapidminer software.

Among the four data mining techniques used to predict the mental health of adolescents based on nutrition, physical activity and leisure time, the random forest technique has the highest accuracy (91.72) and specificity (82.73) and the k-nearest neighbors technique has the highest sensitivity (96.30). In addition, based on random forest techniques, the rule with the highest level of support showed that an adolescent who is in high school, eats breakfast, lunch, and dinner every day, drinks tea and coffee weekly, exercises 2 hours a week at school,also, he has 4 days of physical activity for 30 minutes in the last week, and he goes to school with the service, with 100% confidence has good mental health.

Based on the random forest technique, which has showen the best performance, nutrition has the greatest impact on the mental health of Iranian adolescents. So, it is necessary to think about providing a suitable platform for training parents and adolescents regarding proper nutrition and increasing awareness in the field of adolescent mental health.