But I thought I would include it here for comparison reasons.
It’s an example of Streamlit app where it would be nice to have interactivity between a parent and a child chart.So for example if you zoom on the parent, scatter plot it would be nice if the child, bar chart would update accordingly.
I’ved added an example of a Choropleth Map using data from Our World in Data. Take a look at their site. They have awesome looking charts and dashboards.
In the example I use Bokeh and geopandas to map the spatial data. With this combination you have full control over each polygon you color. Alternatives that also provide control over polygons are HoloViews, HvPlot.
But the dashboard would have been more simple to develop in Plotly or Altair as they have the polygons built in and there is thus no need for geopandas.
I experienced that the file uploader widget can be difficult to use if you allow you users to upload multiple types of files as it does not return any information on the name or type of file uploaded. I’ve posted a feature request Feature Request 896.
I see custom widgets as a way of enabling even more awesome analytics apps in Streamlit.
This hack could be generalized and wrapped into a more general api for creating custom widgets.
So in principle you should be able to get your frontend developer to create custom widgets for clicks on tables, plots and Javascript libraries like Amphion Robotics
Some time in the future the Streamlit developers would develop a secure and robust api for custom widgets making the need for this type of hack obsolete. See for example Issue 428
I think this is a type of app we will see more of in the future. An app which showcases the power of a python package. Just because now its easy to create and it’s so engaging, informative and fun.