At the moment, 11 Excel sheets are read in this way. This is probably not the best way to feed data into the dashboard. A few ideas on how to improve the input performance:
Read Excel files during local build and package the resulting pd.dataframes into a pickle file
I already added three "test file" that show how to make any matplotlib figure interactive, using the bokeh package: sustainableaviation/dashboard/dev/tests. This includes the use of the servable() directive etc.
I would like the data to be "interactive" in the following sense: Mouse-hover to see the point metadata (eg. x, y, label, etc.). Use this application from the panel homepage for inspiration:
For further information on how to get from matplotlib to WASM, compare this issue: