Academic journal article The Journal of Business Forecasting

Answers to Your Forecasting Questions

Academic journal article The Journal of Business Forecasting

Answers to Your Forecasting Questions

Article excerpt

[Q] How can we use simple modeling methods, such as Moving Averages and Exponential smoothing, to forecast for a longer horizon? I note that these methods are good only for forecasting one period ahead. What do we do if we need to forecast, say, twelve periods ahead, using three years of history?

[A] Technically, you are right that we cannot forecast more than one period ahead. To forecast beyond that, we have to make an assumption, that is, sales will continue increasing or decreasing at the same rate. If, for example, sales in the last few months increased, on average, by 5% from one month to the next, we will assume that the same rate will continue in the next twelve months. This is what forecasting vendors do.

[Q] What should be the right forecasting model for the Retail Furniture industry that has a life span of only six months?

[A] The best way to deal with this kind of product is to use its analog, a similar product that was launched in the past. Then determine intuitively whether this new product will do as well, better, or worse than the analog, and by how much. Once you decide on that, the next step will be to break down the total sales by month using the analog's percentages, that is, what percentage of sales on the analog came in the first month, in the second month, and so on. After you have sales data of three months or so of a newly launched product, determine whether sales are coming along as projected, or more than or less than what was projected, and then make adjustments in forecasts accordingly.

[Q] What should be the forecasting model of a manufacturer that makes to order based on tenders. It wants to improve service time/availability by keeping some items in safety stock. The risk is that there is 50/50 chance that a tender may not be won.

[A] To minimize risk, the best thing is to stock some raw material or manufacture limited amount of products without complete configuration. Here I assume you would be able to use the raw material somewhere else if the tender did not come through, and products that are partly configured can be re-configured easily for other customers.

[Q] Our Supply Chain Manager has requested that we provide a forecast accuracy measure that represents a nonweighted average of our products, that is. Mean Absolute Percent Error (MAPE), not Weighted Mean Absolute Percent Error (WMAPE). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.