foreseechange

Weekly Sales Forecasting Model (wsfm)

Operating Instructions

Purpose

This software, developed by foreseechange pty ltd, models historical weekly sales and price data (at least 104 weeks data is required).  Future sales (up to 26 weeks) are forecast based on a user-supplied set of future prices.

This is a common forecasting problem for retailers and vendors.  The model employed is a Box-Jenkins Transfer Function model that has been especially developed and tested for this type of data.  The model quantifies the price elasticity of sales and differentiates between shelf price and temporary price cut impacts.

Installation

The demonstration model should be saved to a directory of your choice, we suggest c:\program files\foreseechange.  It can then be executed using the Start button and then Run.  Use the Browse button to find wsfmdem.exe and then click OK.

The full model is delivered as a self-extracting installation file that installs wsfm on the programs menu.

Operation

First, set up the data series that you want to forecast in a spreadsheet or other program of your choice.  It should look like this:

The weeks should correspond to rows and price should be to the left of sales.  Select the two data columns without headings or week identifiers.  At least 104 consecutive weeks data is required and there must be no missing data.

Switch to wsfm and paste the historical data.  The result should look like this.

Your spreadsheet should also contain future prices for the weeks that you want to forecast.  These should now be copied and pasted to wsfm.  Note that the future prices must be consecutive with the history - ie no gaps.

You can then click the Forecast Sales button.  A chart like this will appear.

This chart indicates how well the model fits the data and shows the forecast sales.  Note that the demonstration version is limited to forecasting up to two weeks ahead rather than 26 and does not provide the diagnostics of price elasticity and TPC%.

Note that the model differences the data so as to detect seasonality and trend.  For this reason, the first 53 data points, while used, are not "fitted".

Price elasticity is a measure of how sales respond to price.  In the example above, the -3.6 price elasticity indicates that a 10% price decrease will lift sales by 37%.  This is a shelf-price elasticity rather than a promotional elasticity.  The TPC% diagnostic indicates the additional impact of a temporary price cut.  In this case, the figure of 20.5 indicates that a temporary price cut increases sales by 20.5% in addition to the price elasticity impact.  The sign of TPC% is usually positive, indicating a positive sales impact of a temporary price cut in addition to the price elasticity impact.

On closing the model fit screen, click the Copy Forecast button to copy the forecasts back to your spreadsheet or other application.

 

If further assistance is required, please contact foreseechange at www.foreseechange.com.