Supply Chain Analytics - Demand Forecasting

Apply naive forecasting, predicting the next month as the previous month’s actuals, to serve as a reference for more advanced methods like moving average, exponential smoothing, or ARIMA.

Investigate moving average forecasting using three- and four-month windows, compare smoothing effects and lag, and note limitations in capturing trend, seasonality, and outliers.

Learn how exponential smoothing differs from moving averages by giving near-term demand more weight using alpha, the smoothing parameter, and how to initialize and forecast with the formula in Excel.

Learn how autoregressive models, as the ARIMA error component, use past time series observations through linear regression to forecast future values, with lag-1 and lag-2 examples and supply chain applications.

Learn to build an autoregressive model in Excel by creating lag one, lag two, and lag three, split data for training, and use regression to evaluate p-values, R-squared, forecast accuracy.

Explore linear regression as a supervised forecasting model, linking sales data to time-based predictors, using y = mx + c and the concepts of dependent and independent variables.

Forecast future demand using a linear regression model in Excel, plot a trend line, display the equation and r-squared, and predict August to November with y = mx + c.

Learn about causal models beyond time series and how multiple linear regression expands forecasting by incorporating historical sales, market share, price, promotions, and promotion spend to improve demand forecasts.

Forecast sales in demand forecasting using a causal multiple linear regression with month, price change, and promotion spend; derive Y = A + B1X1 + B2X2 + B3X3 equation and R-squared.

Use memes to lighten up the learning of supply chain analytics, highlighting how route optimization software can improve delivery performance and easing stress for procurement, logistics, and marketing teams.

Optimize exponential smoothing forecast with Excel solver to minimize MAPE by tuning alpha from 0 to 1. Using alpha 0.4 yields a next-month forecast of 780 with a 34% MAPE.

Align data formats with the forecast horizon, capture promotions, discounts, and Covid disruptions, and apply segmentation by skill, location, and customers to enable organized, accurate demand planning.

Split historical data into training and testing sets (80/20, 70/30) for exponential smoothing forecasts, validate on recent periods, and compare training vs testing accuracy to gauge reliability.

Learn to generate a one-click forecast in Excel using the forecast sheet, based on time-series data with equal intervals and automatic seasonality detection.

Elucidate the limitations of Excel's inbuilt exponential smoothing forecast: limited customization, overreliance on historical data, weak interpretability, and unsuitability for complex models or large data ranges.

Explore the advanced triple exponential smoothing (aaa) method in excel, focusing on alpha, beta, gamma, seasonality and trend; learn how confidence intervals and forecast error KPIs assess accuracy.