Description
The objective of this case is to illustrate forecasting with monthly sales data. It is based on historical data for household goods sales in the United Kingdom from January 1997 to December 2011. Students are challenged to develop a model for this data and to calculate a forecast for 2012 sales based on that model. There is a companion IESE focus case, ADFC-3 "The Jia Wu Expansion," which describes Jia Wu's earlier expansion to Australia. The purpose and the content of the two cases are the same, although the methods used in the present case are more complex. While in the first case, the forecasts were based on a linear trend plus a set of multiplicative seasonals, here we discuss two alternatives: (a) a quadratic trend plus a set of additive seasonals and (b) an additive Holt-Winters model.
This case can be used in many ways: (a) in one session, skipping one of the two models discussed in this note; (b) in two sessions, including both models; or (c) combined with the companion case, using one case for the quadratic trend and the other for the Holt-Winters method. Basic knowledge of Excel or an analogous spreadsheet application is required. The calculations are assumed to be performed using standard spreadsheet functions.
Learning Objective
This case illustrates two forecasting methods for monthly sales data: one based on a quadratic trend and a set of seasonal factors and the other based on the additive Holt-Winters method. Students are expected to have previous experience with linear trends and seasonal patterns. The main points for the class discussion are: (a) how to fit a quadratic trend to time series data, (b) how to calculate additive seasonals for this trend, (c) how to forecast the future values of the series using the trend and the seasonals, (d) how to forecast them using the additive Holt-Winters method, (e) out-of-sample-data validation and (f) parametric vs. non-parametric trends.