«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

清华大学教育研究[ISSN:1001-4519/CN:11-1610/G4]

卷:

期数:

2026年03期

页码:

158-169

栏目:

职业教育

出版日期:

2026-06-20

文章信息/Info

Title:

Predicting Job-Education Match for Higher Vocational Manufacturing Graduates: A Machine Learning Approach

作者:

刘云波¹ 虞梓钰² 杨钋³

1.北京师范大学 教育学部；2.上海交通大学 教育学院；3.北京大学 教育经济研究所

Author(s):

LIU Yun-bo¹ YU Zi-yu² YANG Po³

1. Faculty of Education, Beijing Normal University; 2.School of Education, Shanghai Jiao Tong University; 3. Institute of Economics of Education, Peking University

关键词:

高职毕业生; 制造业; 对口就业; 机器学习; 实习质量

Keywords:

vocational college graduates;  manufacturing industry;  job-education match;  machine learning; skilled talent

分类号:

G717.38

DOI:

10.14138/j.1001-4519.2026.03.015812

文献标志码:

A

摘要:

技能人才是落实制造强国战略的第一资源。在我国制造业长期面临“技工荒”的背景下，高职院校制造类专业的毕业生纷纷离开制造行业，人才流失问题严峻。利用2022年全国13828名高职制造大类毕业生的就业数据，采用梯度提升算法识别出影响毕业生对口就业的27项关键因素。研究发现，实习经历（实习内容与专业的关联性、劳动强度、实习满意度等）的预测效力最强，贡献率接近43%；个人特征（中考与高考成绩、高职阶段学业表现等）、学校办学水平（课程设置满意度、课程与职业标准的衔接性等）以及求职经历均对对口就业有显著预测作用。通过可解释的机器学习方法，揭示了关键变量的非线性影响机制，勾勒出对口就业毕业生的群体画像。研究表明，以实习质量革命为核心突破口，以课程建设和校企合作为主要抓手，构建多主体协同的人才培养生态，可使对口就业率提升10个百分点以上，每年节约教育成本超过6亿元，为建设制造强国和教育强国提供实证支撑。

Abstract:

Skilled talent is the cornerstone of implementing China’s manufacturing power strategy. Against the backdrop of a persistent “skills shortage” in China’s manufacturing sector, a significant number of vocational college graduates in manufacturing-related fields are leaving the industry, leading to severe brain drain. Using employment data from 13,828 higher vocational manufacturing graduates nationwide in 2022,this study employs a gradient boosting machine learning algorithm to identify 27 key factors influencing jobeducation match. The findings reveal that internship experiences (e.g., relevance of internship content to major,workload intensity, internship satisfaction), individual characteristics (e.g., academic performance in middle school and college entrance exams, vocational education achievements), institutional quality (e.g., satisfaction with curriculum design, alignment between courses and occupational/skill certification standards), and job search experiences (e.g., career preferences, number of job offers received) strongly predict job-education match among vocational manufacturing graduates. Notably, internship-related factors demonstrate the highest predictive power (contributing nearly 43%). Using explainable machine learning methods, the study further uncovers nonlinear mechanisms underlying critical variables and constructs a national profile of graduates who achieve a job-education match. Improving institutional operations based on research findings can save approximately 200 to 600 million yuan in wasted educational costs. The conclusions offer significant policy implications for talent cultivation in manufacturing and advancing China’s goals of becoming an education and manufacturing powerhouse.

常用功能

更新日期/Last Update: 2026-06-20