Application of Multi-Layer Perceptron Classification Model to Predict the Professional Direction of Sports Undergraduates Through Personality Traits
Author(s):
Cheng Hua* and Wang Dandan
Background & Aims: Personality traits play a stable and intrinsic role in the process of sport undergraduates coping with the multiple stresses of classroom academic performance and maintaining extracurricular sport. The purpose of this study is to determine the correlation of multilayer perceptron (MLP)models in predicting gender status and major choice among sport undergraduates.
Method: Personality surveys based on the classic Eysenck questionnaire was carried out and MLPs feedforward neural networks with back propagation algorithm were processed by SPSS and cross-validated among the 332 undergraduates. Descriptive analyses and T tests were used to analyze the personality traits of the overall participating subjects. MLP models the original scores of items in the Eysenck Personality Scale were set as covariates, and "gender" and "major" was set to be the predicted output, respectively. Choose the best predictive models from all models.
Results: The personality characteristics of subjects were more extroverted (t =20.838, p =0.000) and more neurotic (t =4.892, p =0.000) and unlikely to be psychotic (t =-0.321, p =0.749). The test outcomes are credible suggested by the Lie score (t =-17.679, p =0.000). The top four items that play an important role in predicting the gender are: N67, N28, E22, E1. The most important items of the E and N dimension scales in the "professional" prediction model are in turn: E85, E1 & N66, N28.
Conclusions:The type of the personality model is ENql, meaning extroverted, neurotic, unlikely psychotic and trusted in the personality characteristics. The application of MLP prediction models is to help undergraduates in choosing their major more easily.