I found this example where filtering on a pandas data frame is used to split data in two and then plot each with a different color vbar.
I want to do something similar, but I am using ColumnDataSource as my data source and, as far as I know, you can’t filter row-wise. Is there a good solution for this that doesn’t require passing my data from ColumnDataSource to a dataframe and back again?
As the CDS is a central data structure to Bokeh, and as there are many possible interactions, the PR is in need of more testing and development to ensure the semantics make sense in all cases. It should be ready and merged for the next 0.12.6 release, however.
However, you may not need that. If you just want to shade different vbars different colors based on some field, you can pass a color mapper as a transform for the fill_color, as show in this example:
I want to do something similar, but I am using ColumnDataSource as my data source and, as far as I know, you can't filter row-wise. Is there a good solution for this that doesn't require passing my data from ColumnDataSource to a dataframe and back again?