Construction of Mental Disorder Prediction Model via Data Mining

Abstract

To explore the psychological barriers of students in college, the prediction model of psychological disorder based on decision tree algorithm and artificial neural network algorithm is established and its accuracy is analysed. In this study, the psychological obstacle prediction models, which on the basis of decision tree algorithm and artificial neural network algorithm, are established respectively, and the two models are used to analyse 1026 records in the psychological census statistics database of freshmen in the Southwest Region University. Through comparing the analysis results of the model with real data, the classification accuracy of the model is verified. Through experiments, the factors that affect the performance of data mining classifier are analysed, and the prediction probability and operation speed of the model are further tested and analysed. The results show that when splitmin value is 39, the classification accuracy of decision tree is 0.98507. The decision tree with 17 leaf nodes has the best classification effect. The probability that the decision tree model and the artificial neural network model are correct in predicting the “normal” of psychological state is 0.987 and 0.9538, respectively; while the probability that the “abnormal” is correct is 0.812 and 1, respectively. The modelling duration of the decision tree model and the artificial neural network model is 0.524s and 7.929s, respectively, and the classification duration is 0.0277 and 0.0275s respectively, which indicates that the models based on decision tree algorithm and artificial neural network algorithm have their own advantages and disadvantages.