TS Forecast factory - Period not accurate

tcrowley128

Hi, I am having an issue with the TS forecast factory generating Periods that match the expected result. I set the TS Model factory to start the time series July 2019 and the TS Forecast factory does not follow the sequential years for the 12 months I'd like to forecast. Below is a screen shot of the TS Forecast Factory output. You can see the years in Period are totally random. Other Groups also have random years like 2142, nothing following a sequential order. This is the input to the TS Model factory before inputting to the TS Forecast factory. Greatly appreciate any help! Tyler

data import.png

alteryx forecast factory error.png

BrandonB

The issue is that you have multiple rows for a given time period record based on the level of grouping that you are trying to run the models on. You have your grouping field set as Country in your TS Model Factory tool, but there are multiple values in the Group column for a given country. For example, row 1 is BMW in Canada for the month of April in 2015. Row 6 is Daimler in Canada for the month of April in 2015. When you feed this in with Country as the grouping field, you have multiple rows for April 2015 for Canada.

If you want to look at a sales forecast on the country level, you need to summarize saying group by month, group by country, group by date, group by year, and sum sales where the "Group" column is disregarded. I think what you might be looking to do is create models not only for the country but also the Group (BMW, Daimler, etc). If this is the case, maybe try creating a new column using the formula tool called Grouping Field and for your formula say [Country]+[Group] to create a new concatenated field. The idea is to then use this field as your grouping field in the TS Model Factory tool.