CODEWITHSUNDEEP
tableau-api
Rob McCauley
Rob McCauley

Reputation: 538

Can Tableau return non-UI results programmatically?

Tableau is an excellent tool for visualizing data. However, it is designed to be the final stop in a data (ETL) pipeline.

My Tableau workbook uses a bunch of Table Calcs to generate a list of "recommended orders". Rather than view these, I want to automate and execute them. This would make Tableau the engine of a quasi-ML process.

In other words, I would like to make Tableau a part of my ETL pipeline and send data to another tier. How can I write a back-end program that executes my Tableau workbook and receives a results dataset?

See the end of this article for example data I want to automate: http://robm26.blogspot.com/2015/10/keep-your-factory-humming-with-tableau.html

Any ideas?

Upvotes: 2

Views: 187

Answers (3)

Stephen Berndt
Stephen Berndt

Reputation: 433

The above answer is correct that tabcmd is the way to pull it out. We use a function in python to generate the tabcmd requests so that they can be batched.

import subprocess

def runTabCmd(cmd):
    # run tableau command and display the output
    print cmd

    if run_tabcmd == 'yes':
        p = subprocess.Popen(
            cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
        for line in p.stdout.readlines():
            print line

You probably already knew that, but for us it was a way to completely automate the pulling and loading into another python package like scikit-learn for a streamlined ML solution

I'm editing this answer to agree with Russell's answer. Tableau is not an ETL tool and should not be used as such. If you absolutely have to do something, you can use what I provided. Otherwise, the best practice is to use a tool designed for the job.

Upvotes: 2

Russell Christopher
Russell Christopher

Reputation: 1707

You're not not going to like the answer I'm going to give you -- "Don't do this".

Tableau isn't meant to be a task in a larger ETL pipeline and the reason you're having problems making it behave the way you want is it's not meant to be done.

Above and beyond the fact that you've figured out how to get a result that you want in Tableau ("the work is done"), Tableau isn't offering you any real value in the scenario you're describing. Use a tool (like Alteryx) that is really purpose built for this sort of work.

Upvotes: 2

bitlamas
bitlamas

Reputation: 742

You can easily use tabcmd to get the results of a view in CSV, which can be used later in your ETL process. If you need to automate it, you can write a script and execute it with a cron job. I, myself, have a few views that are exported to CSV and used later in my ETL stream to feed our CRM.

Just remember to create the view exactly as you want it to be exported to CSV - usually including the order of the fields. Another tip is that I don't let it use the default "Measure Names" and "Measure Values" - to make sure everything is good on my CSV, I have the fields added manually in the row/columns section.

Upvotes: 1

Related Questions