Seasonal Demand Forecasting

[Rashed]

Very recently we have started looking at options for seasonal demand forecasting in business. As such, I have been looking at a few statistical approaches for seasonal demand forecasting such as Bayesian technique, ARIMA, etc. It would be great to know if TM1 is the best technology for statistical analysis for seasonal demand forecasting. If anyone knows of any product which is already built on TM1 then please let me know. It would also be great to know what sort of mathematical models these TM1 based products use.

[belair22]

Hi Rashed

Your question is quite a broad one and quite possibly out of the scope of this forum. The answer really depends on the data you are using (how many rows/tuples, data interaction & complexity of the data).

I've been involved in implementations where Demand Planning/Forecasting was a major part of the development - some reasonably complex modelling going on in these implementations. TM1 doesn't come with its own Bayesian engine algorithm so from a technical point this isn't available 'out of the box' (I'm not sure any BI products would include this for that matter). This obviously doesn't rule out TM1 as an application as the engline could be built in TM1, however I would lean toward a relational database system though I'm basing this assumption on quite a large set of data and quite complex demand planning models.

[rollo19]

If demand forecasting is broken into 1) statistical baseline > 2) demand planning (promotions, new products etc.) > 3) Demand consensus & sales targets > which drive resourcing decisions ..

Can anyone recommend any statistical forecasting packages to prep 1) to load into Cognos Planning/TM1 for 2 & 3?

[rollo19]

I think I can answer my own question: http://www.spss.com/spssetd/

[Michel Zijlema]

I think I can answer my own question: http://www.spss.com/spssetd/

If you're looking for quality statistical analysis and you're not willing to spend a lot of cash you should have a look at the R-Project. Another alternative is PSPP.

Michel