Demand Forecasting

Demand forecasting is the numerical prediction of how many units of a product, service, or capacity will be needed in a future period. Inventory management, production planning, personnel scheduling, and revenue management all rest on demand forecasts — a bad forecast sends the wrong signal to every downstream operational decision. Classical approaches: moving average, exponential smoothing, ARIMA, seasonal decomposition. Modern approaches: regression trees, neural networks, open-source time-series libraries. In practice, three forecast horizons are distinguished: short-term (1–30 days, operational), medium-term (1–12 months, tactical), and long-term (1–5 years, strategic).