Switch default plotting backend to plotly

[LeonieFreisinger]

Status quo Currently, matplotlib and plotly are the two available plotting backends.

Refactoring need For future versions, we plan to only support the plotly plotting backend. In the first step the default should be switched from matplotllib to plotly.

Update

Raised issue with plotly plotting backend : slow processing (rendering) of plotly figures in jupyter notebook and occasionally freezing notebook.

Investigations: Tried to reproduce problem with the following settings:

• RAM: 16GB
• renderer mode: fig.show(renderer="notebook_connected")
• executeduncertainty_conformal_prediction.ipynb in jupyter lab

  
  fig1 = m1.highlight_nth_step_ahead_of_each_forecast(1).plot(
  forecast1[-cutoff:], figsize=figsize, plotting_backend=plotting_backend
  )
  results: execution time 0.11 seconds and RAM used 14.65 GB (that means could not reproduce the issue with my environment setting) 

further infos on plotly renderers: https://plotly.com/python/renderers/

[LeonieFreisinger]

@karl-richter @Kevin-Chen0 unfortunately I could not reproduce the issue of slow processing with my setting. Therefore it would be amazing to have your support :)

Please try the same test, i.e. fig1 = m1.highlight_nth_step_ahead_of_each_forecast(1).plot( forecast1[-cutoff:], figsize=figsize, plotting_backend=plotting_backend ) in uncertainty_conformal_prediction.ipynb and share your execution time, used RAM and where you execture your notebook.

Thanks for the help!