Boosted Model Error

[troyeb4]

Hello,

I'm taking the Udacity Predictive Analytics class and I'm getting the following error when I try to do a Boosted Model.

Error: Boosted Model (21): Boosted Model: Error in getCVgroup(distribution, class.stratify.cv, y, i.train, cv.folds, :

I've made sure there are not nulls and I've reviewed for categorical variables with too many variables.

Any suggestions?

[TrevorS]

Hello @troyeb4,

With Udacity, support is not provided through Alteryx but through Udacity itself.

Please email bizanalyst-support@udacity.com for further assistance.

Thanks!

TrevorS

[Thableaus]

Hi @troyeb4

Could you please post this worflow you're working on?

I'm taking the same course, maybe I could help you out.

Cheers,

[troyeb4]

Trevor,

While the data is coming from Udacity for the class the error is an Alteryx error from the Boosted model. Wouldn't that be something Alteryx could help with?

In the end a mentor with Udacity told me to rebuild the flow to see if it would eliminate the issue and it did. I isolated the boosted model in it's own workflow and no errors.

Thanks,

Troy

[troyeb4]

@thableaus

A mentor at Udacity on the student hub helped me out. I isolated the model to it's own workflow and it corrected the issue.

Thanks,

Troy

[DemandEngineer]

Sadly my problem isn't going away... rebooted... logged into another Desktop and tried the same workflow... error persists.

 I'm getting errors in the Boosted Model but not the other models (Decision Tree, Forest)  using the same Evaluation Data:

"Error: Boosted Model (33): Boosted Model: Error in getCVgroup(distribution, class.stratify.cv, y, i.train, cv.folds, : "

and then an error in the Model Comparison Tool

"Error: Model Comparison (35): Tool #3: Error in match.names(clabs, names(xi)) : "

All predictive variables fields are doubles and I unselected non-demographic data fields. Target Variable is Cluster. Any idea what these errors mean?

[damc]

Hello @DemandEngineer 

Regarding the model error, the Boosted Model tool default configuration uses a 5 fold cross validation to determine the number of trees in the model and I suspect this error is occurring because your estimation sample is too small for it. I´ve changed it to 4 folds and it worked (workflow attached).

I'll try to figure out what´s causing the Model Comparison tool error and I´ll reach out to you if I succeed.

Hope it helps.

Regards