Creating a Forecast with Python that responds to Filters

[kpost]

I have a dataset that consists of several different customers and a summing numeric value associated with each.

My employer wants me to implement forecasting that is responsive to a filter that allows the user to select which customer (or customers) you want to see a forecast for.  This would be easy using the built-in forecasting functionality.... BUT they want the visual to include several lines:  Budget, Actual, and Forecast.  The built-in functionality does not allow a forecast on line graphs that include more than one line.

So what I'm trying to do is implement Holt-Winters forecasting using python, and put the forecast in a new calculated table so that it can be included on the visual along with the other lines.  Is there a way to do this so that it will respond to the customer filter in the same way that the built-in functionality does?  Or is there another, completely different and better way to do this?

Would it be appropriate to just create an extremely complicated measure that implements the Holt-Winters algorithm?

Any and all advice on this topic would be helpful.  Or, if you agree that this should be a built-in capability please at least vote here!
https://ideas.powerbi.com/ideas/idea/?ideaid=bff7fd91-5a2c-467b-b5f3-9a51ba87569b

[DataZoe · Accepted solution]

@kpost You can do this if you create the python as a visual in the report, then use ggplot to create the plot too. Then you can use regular slicers. 

If you want to do python in the power query screen you would just need to have it calculate per customer (and per date, I assume) and then connect back up the output table to the appropriate dims or just include a slicer of that tables customer with your line charts. 

Hope that helps!