Waging a war against how to model time series vs fitting
Matt presented his results at the ISF(http://www.forecasters.org and his presentation is here http://www.forecasters.org/proceedings12/WELLERMATTISF2012.pdf
I see some confirmation of underutilized software tools and real missed opportunities in forecasting!
Take a close look at slide 10,11 and 12.
Also note that on slide 10 that promotions were used 90% of the time. I would have to assume that Matt means "used as an adjustment to the forecast" and part of the CPFR and NOT the baseline forecast as slide 11 clearly shows very few are using causal modeling.
Slide 12 shows that daily data is not being total under used. At the minimum, financial people would want to use that to forecast if they are going to make the month end number and they need to used daily data to do that properly.
Matt saw my post on linkedin.com and wanted to add.........
Indeed Tom, thanks for pointing this out. Your assertions are quite correct: they are not using the kinds of models which can handle these additional variables and tend to rely on judgement to integrate the data.
Now we have in excess of 270 responses the results are more conclusive. I'm just finishing the analysis and preparing the report for all participants. There is still an opportunity to register for the survey, it has not been closed.
I'll share a copy of the results with you Tom in the next couple of weeks, the report will be considerably better than the ISF slides!