An integrated big data framework for labour market demand forecasting and entrepreneurship guidance in vocational education

Skill mismatch and limited personalised entrepreneurship guidance remain persistent challenges in vocational education and training. This study proposes the Big data driven demand and entrepreneurship guidance framework (BD-DEGF), an integrated five-layer decision-support framework that combines job-posting corpora, student academic and behavioural records, macroeconomic indicators, and entrepreneurship-ecosystem signals. The framework applies natural language processing for skill extraction, gradient-boosted modelling for labour-demand forecasting, an attention-weighted Entrepreneurship Readiness Index for student readiness assessment, and a hybrid collaborative and content-based recommendation module for personalised guidance. The framework was evaluated using a composite benchmark dataset parameterised from publicly available sources, including O*NET, ESCO, PIAAC, and GEM. Under these composite benchmark conditions, the framework achieved skill-demand R 2 = 0.923, employability F1 = 0.943, entrepreneurship-readiness F1 = 0.912, and recommendation NDCG@10 = 0.887, outperforming baseline comparators with Holm-Bonferroni-corrected statistical significance. These results should be interpreted as design-validation outcomes under controlled benchmark assumptions rather than evidence of real institutional field performance. The study contributes a formally specified, reproducible, multi-stream framework for vocational-education intelligence and identifies live institutional deployment, fairness auditing, and longitudinal validation as priority directions for future work.