Not able to execute Model Comparison Tool

[abhik24]

I have been trying to execute model comparison tool for comparing three classification algorithms, but the tool keeps giving me the following error. Not able to understand what exactly the error is.

Error: Model Comparison (26): Tool #3: Error in family(object)$linkinv(pred) :
Error: Model Comparison (26): Tool #3: Execution halted
Error: Model Comparison (26): Tool #3: The R.exe exit code (1) indicated an error.

I have attached the workflow here as well.

[fmvizcaino]

Hi @abhik24 ,

Would you be able to package your workflow? That way, I will be able to have your workflow with your macros and input dataset as well.

Best,

Fernando V.

[abhik24]

Attaching the Workflow package.

[fmvizcaino]

Hi @abhik24 

Since you didn't package your model comparisson macro as well, I've configured mine here and it runned with no errors.

Take a look and let me know if yours is still with errors. Attaching mine.

Best,

Fernando Vizcaino

[abhik24]

Hi @fmvizcaino

The workflow package that you have attached is incompatible with my alteryx version 2019.4.8.22007

I am attaching my workflow package again with the macros as well. Please check and let me know.

[fmvizcaino · Accepted solution]

Hi @abhik24 ,

The model comparisson tool needs to work with known information about survivors as it will calculates the accuracy of your models related to what already happened.

So, you only need to change the D anchor of your model comparison to receive data from the T anchor of your filter tool.

Best,

Fernando Vizcaino

[abhik24]

Hi @fmvizcaino 

Thanks a lot. It worked. I was using the F anchor of the Filter tool as the Model Comparison Tool documentation mentions to use Test Data in the D anchor.

Anyway, Thanks again for pointing out the issue. 🙂

[fmvizcaino]

Hi @abhik24 ,

Actually, one thing that I forgot to say is exactly related to that.

When building a predictive model, you need to separate your known dataset in 2.

You can do that with the create samples tool and a standart way of doing it is by separating 2/3 to create your model and 1/3 to validate it. 

2/3 of your T anchor would be entering in all models

1/3 of your T would be entering in model comparisson tool

F anchor would be entering in your best model according to model comparisson tool.

Best,

Fernando V.

[abhik24]

Thanks for the info. Will keep that in mind from now on. 🙂