TITLE:
Analysis of the Match between University Skills and Labour Market Requirements in the Democratic Republic of Congo: A Semantic and Probabilistic Approach
AUTHORS:
Sindani Bukerimanza Moise, Mwiseneza Sebigunda Norbert
KEYWORDS:
Employability, Semantic Analysis, Cosine Similarity, spaCy, Logistic Regression, Higher Education, Labour-Market Requirements, Curriculum Alignment, Skills Mismatch, Democratic Republic of Congo
JOURNAL NAME:
Open Journal of Statistics,
Vol.15 No.6,
December
24,
2025
ABSTRACT: This paper examines the extent to which competencies taught in Congolese universities match the skills required by the labour market. Using official curricula from the Ministry of Higher and University Education and 1560 LinkedIn job postings (2023-2025), we constructed two textual corpora and measured their semantic proximity with spaCy word embeddings and cosine similarity. A logistic regression model was then estimated to predict employability as a function of semantic similarity and academic domain. Results show an overall adequacy rate of 49.14%, meaning that only one out of two university skills is directly relevant to employers’ expectations. The model displays strong predictive performance (Pseudo R2 = 0.6899; AUC = 0.977; Accuracy = 91.8%), and confirms that semantic similarity alone explains about 68% of employability variance, far more than the field of study. Economics and Management and Legal-Political Sciences are the best-aligned domains, whereas Technology, Psychological and Educational Sciences show lower alignment. These findings suggest that curriculum reforms in the DRC should prioritise skill alignment (digital, languages, data, management) over programme expansion. The study offers a reproducible semantic-probabilistic protocol for ministries and universities to monitor skill-labour match.