It's hard to say without a bit more context. If what you want is a pile of standalone HTML files, up front, for each combination, then using output_file/figure/save seems perfectly reasonable. If what you want is to be able to present any of the combinations on demand, but don't need or want HTML files up front, or persisted, then there are a few options:
* use bokeh.embed.components with figure in a flask/django/cherrypy/whatever web app
This would probably work really well if there is some simple REST API scheme that would let you easily create the right plot
* use the bokeh server
This is probably better if you want to have say dropdowns or sliders or whatever that help choose what to show (e.g. like the crossfilter app example), or if you want widget/selection interactions to actually trigger real python callbacks (pandas, scikit-learn, etc) to inform plot updates.
Thanks,
Bryan
···
On Jul 18, 2016, at 10:57 AM, Paul Morrison <[email protected]> wrote:
Hi All,
I'm looking for a best practice on my use case:
1. Read in 2 or 3 csv files into pandas dataframes.
2. Create 20-50 time series plots each with 2 or 3 y values (one for each data frame).
3. Repeating all of the plots after modifying the underlying csv files
Currently, I'm just using a bunch of figure() 's and output_file() but I'm sure there's a much more efficient way. Perhaps something with the server?