Tableau integration

svmayor

I would like to use Tableau to build a visualization dashboard of the RM results. Is there a direct way to interface Tableau with RM? I have looked through the community and haven't found that answer

btibert

Not exactly a direct way, and one that I have not attempted myself, but you could think about your putting your data/exampleSets into the format that you need (presumably tabular) and using a python package such as pantab (which uses the hyper data extract API from tableau) to generate datasets that can be read into Tableau.  

That is probably overly complex, admittedly, and would suggest that you simply write your data to CSV files for ingesting into Tableau separately.

BalazsBaranyRM

Hi,

have you tried the Tableau Writer extension?
https://marketplace.rapidminer.com/UpdateServer/faces/product_details.xhtml?productId=rmx_tableau_table_writer

Using a SQL database is always a good alternative, though. That way, you can use Tableau, optimize the queries by creating indexes and aggregation tables if necessary, and have the data in a common tool if you decide that you'll need additional tools to analyze them.

Regards,
Balázs

btibert

I have struggled with the setup of this widget, to be honest.  I spent a cycle tonight and it's not clear to me what I am doing wrong.  FWIW, there has been some recent dev work on the Tableau side around some packages and APIs.  Most notably, IMO, is the pantab package in python, which is pip installable, and includes the hyper extract API.  Saving a pandas dataframe to a hyper file is a one-liner.

https://pantab.readthedocs.io/en/latest/examples.html

Even if we were forced to have this installed on our local, or via your excellent conda flow, that would be ideal, and would avoid the need to copy files from a DMG file, set env variables, etc.

sgenzer

@btibert - my go-to people for Tableau connections are @M_Martin and @Pavithra_Rao. Maybe they can help here?

If there's a nice Python package, maybe we can wrap it in to a new operator using the Python Operator Framework? @bhupendra_patil ?

Scott

btibert

I have a blog post that I want to writeup today.  Short answer I was able to do this via the Execute Python operator and a conda environment that had the python package installed.  More soon.

M_Martin

Hi Coleagues: The RapidMiner and Tableau Platforms have many great synergies.  RapidMiner's ability to read and write from data in a variety of formats - and run database procedures within a RapidMiner predictice process to prep the data as a prelude to reading data and / or running a model and writing out the results and then reading them into Tableau brings the power of predictive analytics as a service to Tableau users.  At the recent Wisdom 2020 Conference in Boston, I demoed the upcoming RapidMiner / Tableau Integration Component, by which a model deployed to RapidMiner Server (and exposed as a Web Service) can be called from within from a Tableau visualization containing cases to be scored using a Tableau Calculation - the model outputs are then streamed back from RapidMiner Server in real-time to the calling Tableau visualization in a manner similar to way R and Pythin libraries can be called from within Tableau calculations. Keep an eye out for this great Component!  Suggest you stay in touch with Bhupendra Patil re: availability.  Best wishes, Michael Martin 

btibert

It's a user specific solution, but one that can be easily tweaked as a quick workaround.  https://brocktibert.com/post/rapidminer-and-tableau/

sgenzer

This is awesome, @btibert!

SylwiaK

anyone managed to get @btibert solution work on Windows system?

btibert

This should work on any system, the trick is to ensure that you have the proper libraries setup in your environment.  In this case, I am using conda to manage my python environment, and within that env, I have pantab installed.

MartinLiebig

Lets maybe ping @M_Martin here as our expert on tableau.

~Martin