Downsampling Example(s)

I am generating a 14 x 14 pairwise scatterplot matrix where each plot contains 10,000 points so the total number of points would be just shy of 2 million (14 x 14 x 10,000). The grid visualization renders fine using the Bokeh server but it takes a very long time and I know that many of the points overlap. Thus, I’d like to try downsampling. However, I haven’t been able to find any good examples for this (even though version 0.8 is out) and so I am hoping that somebody can provide good representative example(s).

Hi,

Thanks for your feedback. We are working hard on the documentation to include a better user guide that hopefully can help with the server side content. In the meanwhile, have you seen the following examples on the repo?

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/abstractrender.py

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/blaze_source.py

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/serversource.py

Best

Fabio

···

On Friday, May 1, 2015 at 5:13:53 PM UTC+2, [email protected] wrote:

I am generating a 14 x 14 pairwise scatterplot matrix where each plot contains 10,000 points so the total number of points would be just shy of 2 million (14 x 14 x 10,000). The grid visualization renders fine using the Bokeh server but it takes a very long time and I know that many of the points overlap. Thus, I’d like to try downsampling. However, I haven’t been able to find any good examples for this (even though version 0.8 is out) and so I am hoping that somebody can provide good representative example(s).

Thanks for the links, Fabio. Can you tell me if I need to do anything special (I’m trying the abstractrender.py example) and have never used blaze before (let alone along with bokeh-server)?

···

On Tuesday, May 5, 2015 at 8:25:11 AM UTC-4, Fabio Pliger wrote:

Hi,

Thanks for your feedback. We are working hard on the documentation to include a better user guide that hopefully can help with the server side content. In the meanwhile, have you seen the following examples on the repo?

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/abstractrender.py

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/blaze_source.py

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/serversource.py

Best

Fabio

On Friday, May 1, 2015 at 5:13:53 PM UTC+2, [email protected] wrote:

I am generating a 14 x 14 pairwise scatterplot matrix where each plot contains 10,000 points so the total number of points would be just shy of 2 million (14 x 14 x 10,000). The grid visualization renders fine using the Bokeh server but it takes a very long time and I know that many of the points overlap. Thus, I’d like to try downsampling. However, I haven’t been able to find any good examples for this (even though version 0.8 is out) and so I am hoping that somebody can provide good representative example(s).

Aha, based on this previous post:

https://groups.google.com/a/continuum.io/forum/#!msg/bokeh/vM8erulLZ7o/EXnZw87gi6YJ

I needed to do three things:

  1. install multiuserblazeserver (conda install -c https://conda.binstar.org/bokeh multiuserblazeserver)

  2. install abstract_rendering (pip install abstract_rendering)

  3. update blaze (conda update blaze)

After this, then the abstractrender.py example worked (after firing up the bokeh-server from the command line).

···

On Tuesday, May 5, 2015 at 9:55:15 AM UTC-4, [email protected] wrote:

Thanks for the links, Fabio. Can you tell me if I need to do anything special (I’m trying the abstractrender.py example) and have never used blaze before (let alone along with bokeh-server)?

On Tuesday, May 5, 2015 at 8:25:11 AM UTC-4, Fabio Pliger wrote:

Hi,

Thanks for your feedback. We are working hard on the documentation to include a better user guide that hopefully can help with the server side content. In the meanwhile, have you seen the following examples on the repo?

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/abstractrender.py

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/blaze_source.py

https://github.com/bokeh/bokeh/blob/master/examples/plotting/server/serversource.py

Best

Fabio

On Friday, May 1, 2015 at 5:13:53 PM UTC+2, [email protected] wrote:

I am generating a 14 x 14 pairwise scatterplot matrix where each plot contains 10,000 points so the total number of points would be just shy of 2 million (14 x 14 x 10,000). The grid visualization renders fine using the Bokeh server but it takes a very long time and I know that many of the points overlap. Thus, I’d like to try downsampling. However, I haven’t been able to find any good examples for this (even though version 0.8 is out) and so I am hoping that somebody can provide good representative example(s).