Profit From Your Forecasting Software

CHAPTER 8Automation and Choice

8.1 INTRODUCTION

Even if you are using the most sophisticated forecasting software, you will still have to rely on your judgment to make some decisions. You will need to decide how much sales history to feed into the computer – often, companies retain only short sales histories, which will restrict the range of forecasting methods that can be used. You will also need to decide how sophisticated you want the forecasting method to be. The forecasts of more sophisticated methods may lack credibility amongst managers who do not understand their rationale. Alternatively, their lack of transparency may cause people to have unwarranted faith in their forecasts because they believe that complexity equates to accuracy. Complex methods may also require the collection of costly extra data. If you are using a method like multiple regression, you will also need to decide which predictors might be potentially useful in forecasting sales. For example, you may decide to have a list of candidates that includes advertising expenditure, price, competitors' action, and weather variables, but you will also need to consider whether you can accurately predict these for the period you are forecasting.

This chapter will provide guidance when you face these sorts of choices. For example, we will look at how much past data you need to apply different forecasting methods and ask whether more complex forecasting methods tend to be more accurate.

While there is always a need ...