Large time series, bokeh-server and iPython Notebook


I’ve recently started using iPython Notebook as a way to record and present my research however I’m having a couple of problems. My research involves taking large time series measurements in matlab which I then load to my notebook using loadmat from scipy. Ideally what I would then have is some type of interactive plot of the data which allows me to zoom, pan, select points etc. in the same way as a matlab figure would allow.

I tried using Bokeh however as other people have found it is very slow with large data sets and bokeh-server is required. I’ve come across two things that seem particularly relevant, the first is this video: and the second is this answer: Is the described downsampling server what is being used in the video and does anyone have an idea of when this will be meshed with the standard bokeh server?

Also is there a way to embed the bokeh-server plots within the notebook that created them? All the example .py files open a new tab at http://localhost:5006/bokeh but embedding the plots is one of the main attractions of iPython Notebook for me.

If a simple way to dynamically downsample doesn’t exist yet and if it does exist but can’t be embedded in a notebook could someone suggest another plotting tool which would allow me to embed an interactive large time series plot in a notebook?

Thanks for your help,