Python Script in Power BI
It's great that Power BI already has the capability to run R scripts for data transformation. It would be even better if we could run Python script.
Alan Pham commented
I want to function run script python in PowerBI Server
Matthew Bitter commented
on prem data gateway refresh for python please.
Please add the on-premise gateway refresh for Python script data source asap.
It's a huge productivity gain for report developers and data engineers.
Any updates on this yet?
Python has basically _saved_ my dataset, but I want to share it with co-workers. As of today, I need to refresh and publish the dataset on my local machine and share it manually every day to my co-workers. This is a hazzle.
It does not look like we have the ability to schedule refresh using the On-Premises gateway. Is this a feature in the pipleline?
Christina Chen commented
Is it available in Power BI Desktop for Power BI Report Server on Prem?
That would be nice!
Karl Duckett commented
Keep updating the Python functionality within Python! It's super power and will only benefit Power BI's dominace!
Karl Duckett commented
Keep updating the Python functionality within Python! It's super power and will only benefit Power BI's dominance!
Robert W commented
I recently had the opportunity to try out the new Python script as a source feature. I was impressed to see Power BI identify every DataFrame in my script and display them as a table selector. The PythonScriptWrapper to export to csv and then import is clever.
The one issue I have with the feature is how Power BI handles refreshing each table from the single script. Even though the tables were originally imported together, they are no longer related after that point. Each table will execute the script when a refresh is requested. This is not only very CPU and memory intensive when parallel refresh is enabled, but it takes much longer than the original import.
Can this behavior be changed in order to run the script once to refresh all of the tables generated by that script?
I am very lucky that this feature already is in preview stage.
Been testing a little bit, and i could use some of my most used python scripts with only little modification.
Most of them were running fine.
Good job and a big "Thank you"!
B. Buil commented
After executing the Power BI update of 16-09-2018, the whole system was crashing time after time.
I have removed the latest updated version of 16-9-2018 and I have installed the old version of 1-9-2018 back again and the system was stable as before…...
It would be great if we could access all tables in our data model to create custom measures in Python. At this point it seems that you can ETL one table at a time in the import phase, and create visuals in the visual element. If we can write python functions to act as DAX functions I think this will be very useful.
Arne Petter Dragest commented
@Peter, The documentation still seems a bit sparse. You'll need pandas and matplotlib to make Py visualizations. If you're running Anaconda you should be fine. There seems to be an issue with locale settings if you're based in a country where comma as decimal separator is the norm. So if you install Anaconda and check your locale settings, you should be able to make the same chart that is shown in the Power BI desktop august feature summary.
Python visuals are currently supported only in Power BI Desktop. I tried putblishing to Power BI on the web, but I only got that exact error message with a link directing me to a page about R visualizations. So as I said, the documentation seems a bit sparse.
Check out these resources:
Some issues to be aware of:
Link in the error message from Power BI services:
Is there any documentation on this? Like a list of packages supported? Do the visuals work in the service?
The preview worked great for me! One thing that would make it easier to use would be to enable the drop-down ability like in a python editor i.e. when I type "pandas." I get a drop-down list of the pandas functions. Thank you so much for creating this capability!
Thomas LeBlanc commented
I am getting an error no module named 'pandas'
enhorabuena.... mucho esperar.... ahora la documentación para sacar el máximo provecho.... felicitaciones al equipo.. un humilde usuario
will this support something like Bokeh?
cool! thank you
Chris Zumbika commented