R - Can you set up a R Model for later use by test cases to minimize the time to run....

[kconner]

Hi,

I am designing an App that utilizes a Random Forest & Nearest Neighbor Model to give a predicted Net Revenue value & nearest comparisons for some input data.

The model works great (yay Alteryx), but I am wondering if I can 'separate' the model part and the input data testing to improve performance as it takes about 3 mins now.

What I would like to do is similar to traditional R modeling where I make the model and run it. Then with the input data, I just call it against the model and the result is calculated in a minimal time. Is there a way to store this already run Forest Model for a quick run later?

Thanks,

[MarqueeCrew · Accepted solution]

@kconner,

Absopositively!  When you build the model, you can output the model to a yxdb file.  Now you've got a static model.  When it comes to scoring within the next workflow, you can input the model object (yxdb) and proceed to score the data without reconstructing the model.

Cheers,

Mark