Calcite | Level 5

## Sales Forecast

I would like to forecast Sales, using workdays. Example:

I haves Workdays of previous months and the sales values, now I want to forecast, November. I know the amount of workday that November has and would like to use it to forecast the final sales.

Is it possible? What is the method? I would like to have a syntax example, if possible .
Thanks!

 MÊS workdays sales jul-18 20 110,4 ago-18 22 105,6 set-18 19 93,7 out-18 21 96,1 nov-18 20 ?????
4 REPLIES 4
Ammonite | Level 13

## Re: Sales Forecast

IS MES dates?  if so then those are typo's.

@Lyev wrote

 MÊS workdays sales jul-18 20 110,4 ago-18 22 105,6 set-18 19 93,7 out-18 21 96,1 nov-18 20 ?????
Calcite | Level 5

## Re: Sales Forecast

MÊS= MONTH,it is a date.
Calcite | Level 5

SAS Employee

## Re: Sales Forecast

If I'm understanding your question correctly, you have 4 data points from the previous 4 months, and you want to forecast one period out, is that right?

If this is the case, please keep this in mind first: all forecast are wrong, but some are useful.

A simply model to try is regression on the number of workdays. This is under the assumption that only workdays affects the sales, trend or seasonality or other factors are not important in sales forecast.

If you think trend is more important than the number of workdays, you can ignore the number of workdays, just use simple smoothing models on the sales record.

With only 4 data points, there is nothing much you can do, other than 1) figuring out what is the most important factor that affect the sales forecast, and 2) applying a simple model that incorporate this factor.

However, if you have a lot more data, I think ARIMAX model is the way to go. It not only considers the time series related feature (ARIMA), but also incorporate the input variable (number of workdays). If you have license for forecast server, you can use PROC HPFDIAGNOSE and PROC HPFENGINE. If you have Visual Forecasting, you can use PROC TSMODEL with ATSM package. Both approaches will automatically identify the best ARIMA model for you. If you don't have the license above, you can use PROC ARIMA too.

Discussion stats
• 4 replies
• 1017 views
• 1 like
• 3 in conversation