Machine learning prediction model based on enhanced bat algorithm and support vector machine for slow employment...

Area of Science:

• Computer Science
• Machine Learning
• Data Mining

Background:

• College student employment is critical for national development and social stability.
• Increasing graduate numbers, employment pressures, and the epidemic exacerbate 'slow employment' challenges.
• Data mining and machine learning offer solutions for graduate employment prediction and guidance.

Purpose of the Study:

• To develop an accurate and interpretable model for predicting college graduate employment prospects.
• To address the growing issue of 'slow employment' among recent graduates.
• To provide effective employment guidance for universities, governments, and students.

Main Methods:

• Proposed a feature selection prediction model (bGEBA-SVM) combining an enhanced bat algorithm with a support vector machine.
• Utilized Gaussian distribution-based and elimination strategies to optimize feature selection for improved efficiency and accuracy.
• Trained and tested the model on a dataset of 1694 college graduates from Zhejiang Province in 2022.

Main Results:

• The bGEBA-SVM model achieved a prediction accuracy of 93.86%.
• Identified key factors influencing employment outcomes, including further education, student leader experience, family situation, career planning, and employment structure.
• Demonstrated superior performance compared to peer and well-known machine learning models.

Conclusions:

• The bGEBA-SVM model is a high-performing and interpretable tool for graduate employment prediction.
• The findings offer valuable insights for improving graduate employment services and strategies.
• The study provides a feasible solution to alleviate the 'slow employment' problem.