Alteryx Designer Discussions

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

What drives the big difference between using assist modeling tool and predictive tool?

Gualigee
8 - Asteroid

I used predictive tools: linear regression and random forest to predict. 

With the same source data, the same target and predictor variables, assist modeling tool generated a big different result (16% variance for LR). 

 

I understand one uses R, the other uses Python. Thus, the results cannot be exactly the same. 

However, the statistical theory behind the scenes should be the same. 

I wonder what drives the big difference between using assist modeling tool and predictive tool?

 

I have attached the workflow for your reference. 

 

Thanks

 

2 REPLIES 2
AkimasaKajitani
15 - Aurora

Hi @Gualigee 

 

Please check the hyper parameter.

For example, Linear Regression of Intelligent suites is set as Fit Intercept is On and Normalize is On.

AkimasaKajitani_0-1642245292548.png

 

But the Linear Regression of R based tool is set as all off.

AkimasaKajitani_1-1642245395298.png

 

If you use the same hyper parameter between R based tool and intelligence suites, the difference will be smaller. 

 

 

Gualigee
8 - Asteroid

Thanks, @AkimasaKajitani it makes sense now. You rock!

Labels