Harry Poku

Harry Poku is an MSc Data Science student at the University of Ghana and a Graduate Researcher at DRAKA Labs with a professional background in Information Technology and enterprise systems.

His research focuses on the intersection of Graph Neural Networks, Social Network Analysis, demand forecasting and applied data science, with particular interest in developing intelligent models that capture relationships and dependencies that conventional predictive approaches may overlook.

His current MSc research investigates an important limitation of traditional product-demand forecasting: products are frequently modelled as independent observations even though real purchasing behaviour creates relationships among products. Harry investigates whether these relationships can be explicitly represented as networks and incorporated into machine-learning models to improve demand forecasting.

His research combines Social Network Analysis with Graph Neural Networks, providing both interpretable analysis of product relationships and advanced graph-based predictive modelling. It uses public benchmark transaction data together with anonymised industrial FMCG data, examining graph-based forecasting methods in both experimental and real-world environments.

Beyond predictive accuracy, his research places emphasis on interpretability, robustness and practical relevance, particularly within FMCG supply chains where forecasts influence inventory, procurement, distribution and commercial decisions.

At DRAKA Labs, Harry contributes to emerging research in graph machine learning, relational intelligence, forecasting and supply-chain analytics.