ALTERYX INSPIRE | Join us this May for for a multi-day virtual analytics + data science experience like no other! Register Now
The Alteryx Community will be temporarily unavailable for a time due to scheduled maintenance on Thursday, April 22nd. Please plan accordingly.

Alteryx Designer Discussions

Find answers, ask questions, and share expertise about Alteryx Designer.

Generate Forecast where data has duplicate dates

5 - Atom



I am looking to generate and compare forecasts using the ETS and the ARIMA models where my dataset is at a slightly disaggregated level. This means that my data set would have duplicate date entries. Please see example below:


I would like to generate a forecast for 1A, 1B, 2A, 2B and 3B. I noticed that Alteryx gave me an error when I had my data set up this way. Can you help me with the methodology I can employ to generate a forecast at this level?





Example Data Set

Attribute 1Attribute 2DateHistory
1A2 Feb 201710
1A2 Mar 201720
1A2 Apr 201725
1B3 Feb 201720
1B3 Mar 201720
2A2 Feb 201720
2B3 Mar 201710
3B3 Apr 201730
12 - Quasar


This workflow segment will put the data into a format that the forecasing tools can consume.



12 - Quasar


I just noticed that my previous yxmd file had a setting error for handling dates.  Please use this version instead.

5 - Atom

Hi there - this is very helpful. However, with the real dataset I might end up having close to 2,000 Attribute 1 and Attribute 2 combinations which might make it difficult to incorporate into the flow. Is there any way to "group" attribute 1 and 2 prior to generating the forecast?

12 - Quasar


Here is a variant on the previous wworkflow that may suit your needs.  You will need to work up your own test data to see that it scales correctly.