An Adaptive Model Selection Framework for Demand Forecasting under Horizon-Induced Degradation to Support Business Strategy and Operations

The project is a nascent academic reference implementation (8 days old, 0 stars) for a paper addressing the classic problem of model selection in demand forecasting. While the focus on 'horizon-induced degradation' is a valid niche in supply chain management, the project currently lacks any adoption or ecosystem. From a competitive standpoint, this space is heavily crowded by both open-source heavyweights (Nixtla's StatsForecast/NeuralForecast, Unit8's Darts, and AutoGluon-TimeSeries) and cloud providers (AWS Forecast, Google Vertex AI Forecasting) that already implement automated model selection and ensembling (AutoML). The defensibility is low because the 'secret sauce' is an algorithmic approach that can be easily replicated or integrated into more established libraries. Frontier labs like Google (with TimesFM) and Amazon (with Chronos) are moving toward foundation models for time series, which may eventually render complex model-selection heuristics for legacy statistical models obsolete. The high platform domination risk stems from the fact that major cloud providers view demand forecasting as a core enterprise service and frequently update their AutoML wrappers to include the latest academic selection strategies.