Machine Learning-Based Forecasting for Industry 4.0 Skill Demand: Evidence from Ethiopia's Garment Sector

Ethiopia's "Digital Ethiopia 2030" objective is beset by persistent structural skills mismatches as TVET institutions find it difficult to link technical industry demands with pedagogical results. at order to simulate the supply-demand dynamics of Industry 4.0 skills, this study looks at digital leadership and institutional preparedness at polytechnic institutions. The study assesses institutional constraints by combining a quantitative cross-sectional survey (N=89) analysed using descriptive and regression statistics with a time-series machine learning technique (Random Forest vs. Linear Regression in WEKA). The results show that while Random Forest forecasting greatly outperforms baseline linear models by lowering Mean Absolute Error (MAE) by 63%, digital leadership dramatically reduces institutional inertia. The research provides useful short-, medium-, and long-term frameworks for TVET workforce planning, including digital lab upgrades and trainer certification. This study builds on the Human Capital Theory by showing that predictive analytics and digital infrastructure are crucial elements of modern human capital. By integrating algorithmic forecasting with institutional readiness, the study offers an objective way to replace legacy planning and close the gap between business and academia. By showing how vocational institutions may become adaptable centers of innovation by transitioning from static administration to predictive algorithmic governance, this study ultimately advances knowledge and directly supports systemic socioeconomic progress and sustainable industrial growth.