Browsing by Author "Nakandi, Proscovia Norah"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Decision Support Tool for Optimal Crop Planning: Bugiri Sugarcane Farm
    (Busitema University, 2026-07-13) Nakandi, Proscovia Norah
    This study developed a data-driven Decision Support Tool (DST) for crop yield prediction and resource optimization using computational optimization techniques to address agricultural challenges in developing regions such as Uganda, particularly climate variability, inefficient resource use, and limited access to predictive technologies. To ensure maximum data integrity and localized relevance, the framework was trained using historical climate datasets cross-validated with ground-truth records from the Uganda National Meteorological Authority (UNMA), correcting for satellite retrieval biases Focusing on sugarcane production in the Busoga sub-region of Eastern Uganda, the framework integrated geospatial analysis, climatic modeling, soil hydraulic characterization, evapotranspiration simulation, and intelligent optimization within a unified system. The tool’s functionality was specifically enhanced by incorporating location-specific soil nutrient parameters, including Nitrogen, Phosphorus, Potassium, and pH levels, to guide objective varietal selection. Historical climate, soil, and agronomic datasets were used to train a Hybrid GRU-LSTM deep learning model and an Extreme Gradient Boosting (XGBoost) model, both evaluated using R², NSE, RMSE, and MAE, with the GRU-LSTM model outperforming XGBoost achieving a training Nash-Sutcliffe Efficiency (NSE) of 0.9986 and a testing NSE of 0.8212. To transition these complex models into a practical tool for smallholders, the research prioritized the design of an intuitive Graphical User Interface (GUI), ensuring the framework is accessible and providing actionable "Farm Action Plans" for non-technical users in rural Ugandan contexts. To convert predictions into actionable decisions, the system incorporated Analytical Hierarchy Process (AHP), Non-Linear Programming (NLP), and Genetic Algorithm (GA) optimization for crop selection, irrigation scheduling, and resource allocation, resulting in improved Water Use Efficiency (WUE) and Irrigation Water Use Efficiency (IWUE), including over 20% irrigation water savings without significant yield loss. Overall, the DST offers an intelligent, science-driven platform for enhancing yield forecasting, irrigation planning, and economic decision-making, supporting climate-smart agriculture and sustainable productivity in Uganda and comparable tropical systems.