InterReg Tutorial: Create a Sales Forecast

Sales forecast

 

Given a table of sales data, in MS-Excel format, as a CSV-file or from another data source. This table has the columns 'Date' and 'Sales Volume' of single sale items.

Step by Step Instructions:

 

1. Create a new graph

Choose 'new graph' from the tool bar

  1. Data type of the x-axis: Date and Time
  2. X-axis format: Year / Month
  3. Y-axis format: Currency
 

2. Data Import

Choose 'Data->Import sample points' from the main menu.

  1. Select a data source and click on 'open'
  2. Connect with the data source
  3. In the following db-Import dialog, assign the values to the target fields 'X-Value' and 'Y-Value'. X-values should contain the date/time values, Y-values the sales volumes (floating point number).
  4. Click on 'Import'.

A new dialog with import options shows up

  1. Check 'Group X-Values, cumulate Y-Values', 'per month'
  2. Leave all other options at their defaults
  3. Click on OK to start the import.

When the import is finished, all sample point should be displayed as in the screen-shot above.

Delete the last (right most) sample point, when it belongs to an incomplete month. Such incomplete data would tamper the result. (Click on the sample point to select it, then hit the Del-key)

 

3. Calculation of the Trend

In the left side bar menu click on 'Tasks->Statistics->Trend'.

The program should draw the trend line straight through the sample points.

 

4. Calculation of the sales forcast (integral)

Click on the left side bar menu on Tasks->Statistics->Integral->Simpson.

  • For the 'graph' select our earlier created 'Trend'
  • Select 'Integral between Graph 1 and X-Axis'
  • Enter the desired interval for the forecast, e.g. 'from 1/1/2015 to 1/1/2016'
  • Select 'Step-size of the values: Month' (because our values have been cumulated by month)
  • Click on OK to run the calculation.

The program draws the integral as a colored plane in the given interval between the trend-line and the x-axis. It will then show the calculated value based on the extrapolated trend. The result should look somehow as in the screen-shot above.

You can later see the exact result, when you double-click on the title in the curves list at the bottom of the sceen.

You can continue and calculate more forecasts for other intervals.

 

Done !