R predictive tools vs python tool in Alteryx

[Hagar-Usama]

Hello,

If I would create a ML model, which is best, R predictive tools or python tool (in Alteryx)? Could I integrate both to work together?
What are the limitations of both?

Thanks!

[mjf]

The R Tool and Python Tool are both developed with a very simple idea: Alteryx develop the ability to interact with other tools, the user inputs the R/python code as they see fit. In terms of ML, R isn't "better" than python, and python isn't "better" than R. Like all programming languages, the biggest factor is how you write the code. There are plenty of ML topics on R and python, I'd say more for python. There are also plenty of native Alteryx tools that are built using R for ML purposes, and other predictive purposes. Look at all tools with a little R symbol in the bottom-left corner, right-click and open macro for further information... or look at the help pages on the tools.

In terms of limitations, it is really down to the user. If you find a R predictive tool is limited in a certain way, then open the macro and copy the code to a separate R tool and amend the code as you see fit.

For an extra fee, you can use ML tools in Alteryx's "intelligence suite", which utilises the python code in their github repositories.

[Hagar-Usama]

Thanks @mjf for your insights

I just have a couple of questions,

- is Alteryx opensource?

- Can I integrate R predictive tools with python code? does it make sense?

[mjf · Accepted solution]

No, Alteryx is not open-source.

I think your request makes sense. Alteryx's predictive tools, which are built with R, work like any other tool in that the output from one can feed into another; Alteryx have kindly given these functions in the R Tool/Python Tool configuration windows. So, you can have the output from a python tool as input into a predictive tool with R. To run a predictive tool within a python tool will be unnecessarily complicated, certainly far more trouble than it is worth... and I can't see it working anyway. Although I've never tried it. You'll need a python package that can execute R commands and you'll need to work with the Alteryx R packages before calling the predictive tool. It sounds messy and it sounds like it won't work. If you need to use python and predictive tools in R, then just divide your python code into multiple python tools with interconnecting predictive tools in R. That way, your python code will behave like a wrapper to R predictive tools. I hope that makes sense.

Good luck.