Hello everyone !
After trying Streamlit for building a quick prototype to present some analytics results internally, I decided to spend more time with the library
A year ago I built a little interactive regression using ipywidgets inside a Jupyter notebook. The goal was to show in a ML introductory talk how we could manually search for the best parameters to fit a function/decision tree/etc… to data.
I thought this would be a good app to port to Streamlit, so I quickly made a first draft with sliders in the sidebar updating the Altair chart for the function with selected parameters, so far so good !
then realized that unlike ipywidgets, we can’t restrict updating the execution to mouse release events, at least at first glance of the API reference.
I think the pros & cons for continuous update are well documented in the ipywidgets disabling continuous update doc, so I think it could be interesting to provide the user with the choice between continuous update, no update and have another button rerun the script, and restrict execution to mouse release events.
It is something in the works, or maybe others have encountered this and found other solutions ?