Prediction Method of Higher Education College Students’ Employability Based on Data Mining

Conventional methods for predicting the employability of college students in higher education mainly use ROC (Relative Operating Characteristic) to generate employment prediction framework, which is easily influenced by multi-prediction sensing, resulting in poor prediction performance. Consequently, there is a need to develop an innovative approach for forecasting the job readiness of university students within the realm of higher education, utilizing data mining techniques. The use of XGBoost and other algorithms to classify the feature categories of college student employment ability prediction data ensures the comprehensiveness and systematicity of feature selection, and reduces the bias caused by subjectivity. Based on this, using data mining to construct a prediction model for the employment ability of college students, further improving the generalization ability and anti-overfitting characteristics of the mining model. Design a prediction index system for college students’ employability, optimize the mining effect, and complete the prediction of college students’ employability. The results of case analysis show that the MSE and MAPE of the designed forecasting method for college students’ employability in higher education are low, which demonstrates that the devised predictive technique for assessing the employability of college students exhibits a strong forecasting capability, reliability, and holds significant practical value. It has also contributed to advancing the optimization and evolution of the societal talent framework.