Forecasting IT Skill Demand in the East African Labor Market: A BERT-Based Analytical Framework

This research addresses the critical misalignment between IT skill supply and demand in East Africa by proposing a novel forecasting framework based on a Bidirectional Encoder Representations from Transformers (BERT) model. Existing labor market information systems often provide retrospective analyses, leaving policymakers and educators without the foresight needed to adapt curricula and training programs proactively. To bridge this gap, this research develops and validates a predictive analytics framework designed to forecast IT skill demand. The methodology applies a contextually fine-tuned BERT model to perform semantic analysis on unstructured data from regional job markets. This enables the precise extraction and categorization of IT competencies from natural language text. The framework is designed to generate predictive insights into future skill requirements. The goal is to provide timely, data-driven intelligence to directly inform education policy, curriculum development, and targeted upskilling initiatives, thereby supporting a more responsive digital workforce in the region. The forecast findings show Excel with a 0.05x multiplier growth (slow growth, though current demand is high), while Java and JavaScript have higher demand. The jobs distribution charts show web development at 43 % and Cloud/DevOps at 41 %. By generating granular, forward-looking intelligence, this framework equips educational institutions, training providers, and government bodies with the evidence base required to align skill development with the trajectory of the digital economy, fostering a more responsive and competitive workforce in the region.