College students' mental health evaluation method based on WLLE and MLP
Jiao Ding, Li Yang · 2021
To address the problems of poor generalization performance, high computational cost and low prediction accuracy of existing methods for evaluation of college students' mental health, we propose a mental health evaluation method based on WLLE (weighted locally linear embedding) and MLP (multilayer perceptron). First of all, by using the cluster sampling method, with SCL90 as the test scale, the mental health data of college students was collected using paper questionnaires and online questionnaires, and the data was pre-processed. Then, the WLLE algorithm is employed for dimensionality reduction of high-dimensional data, and the key features of dataset are extracted. Finally, the nonlinear classifier MLP is used to classify the extracted feature data, and the mental health states of samples to be tested are evaluated. The experimental results prove that compared to the existing mental health early warning methods, the method proposed in this paper can provide higher prediction precision and stronger generalization performance.