Macros & Saving individual predictive model

[stevehughes]

A question for the good and wise. Well, wiser than me which isn't too hard.

I've finally, with much appreciated assistance from the community, got a batch macro working that loops through a list of products and builds a predictive model for each. I'm trying to save, using the output tool, each model individually for use in scoring later on. Issues encountered, (i) the output node cannot see "SKU_ID" as a way to dynamically rename the output file, (ii) it just creates a single alteryx file for all SKUs. This fails because I might want to train a model for a new SKU without impacting the other ones.

So, what might I be doing wrong? And also, any suggestions for place to look in terms of scoring? Presumably , it needs to loop through each saved model output, grab the product ID from the name, and somehow pass that through the scoring process.The current process "looks" like its just creating a huge cross join when I add the saved model(s) and new scoring data.

As always, many thanks.

S

[BrandonB · Accepted solution]

Can you use the option in the output data tool to take the file/table name from field? You can have each loop of the macro saving off the model into its own Alteryx YXDB file. 

 

[stevehughes]

Brill. I was trying to get that to work for a while.Whatever I did I couldn't see any fields available in their.Seemed as I wasn't using SKU_ID in the predictive model it wasn't getting past that stage.However, adding another action based on the SKU_ID variable got it to appear.

Many Thanks!