ML Demand Forecasting for Essential Medicines at Joint Medical Stores (Uganda): Data, HARP Pipeline, Models, and Results

{ "title": "ML Demand Forecasting for Essential Medicines at Joint Medical Stores (Uganda): Data, HARP Pipeline, Models, and Results", "upload_type": "dataset", "description": "Complete reproducibility repository for a master's dissertation on machine-learning demand forecasting for essential-medicines supply at the Joint Medical Stores (JMS), Uganda. Contains the raw JMS consumption workbook (2020-2024, no patient data), the WHO EML and Uganda EMHSLU reference vocabularies, the HARP five-stage harmonization pipeline (code + audit registers + harmonized panel), the three forecasting models (Random Forest, ANN, Linear Regression) with results and diagnostics, and a leave-one-stage-out ablation quantifying the harmonization layer's contribution to forecast accuracy. All figures reproduce from the deposited code.", "creators": [ { "name": "Wyclif, Kaye", "affiliation": "[Add your institution]" } ], "keywords": ["demand forecasting","pharmaceutical supply chain","data harmonization","entity resolution","random forest","Uganda","essential medicines","reproducibility"], "license": "other-open", "access_right": "open", "notes": "Raw JMS data are aggregate (product-branch-year) with no patient-level information. If publishing openly, confirm redistribution terms with the Joint Medical Stores; consider Restricted access for 02_data if required." }