Geospatial applications using datashader, hvPlot, and GeoViews

Showcase
, ,
1

Hi there,

these are from the end of last year but maybe still interesting for some folks who are into geospatial applications of HoloViz tools.

The first example simply repurposes the NYC Buildings use case to work with OSM data.

image
image841Ă—528 68.6 KB

The second example uses hvPlot to visualise a time series of supervised classifications.

image

A more elaborate piece of work for my MSc thesis compared different geospatial Python libraries for visualising large vector data, with datashader and hvPlot doing extremely well along many indicators. You can find the results on this repo.

Let me know what you think! Still got plenty to learn :slight_smile:

Best,
Gregor

3 Likes
2

I love https://github.com/gregorhd/mapcompare! It would be nice to promote that more.

I wonder why GeoViews in some cases was taking many times longer than hvPlot, when hvPlot is simply a wrapper on top of GeoViews…

3

Hi Jim, thanks a mil! Will PM you about our plans for getting it wider publicity.

Yes, the GeoViews results kept puzzling me as well. I would have also expected hvPlot to take longer than GeoViews…

4

Did geoviews get used in the hvplot benchmark e.g. geo=True / coastline=True?

Also, maybe the version of cartopy / geoviews is different?

1 Like
5

Yes, geo=True (coastlines=True not needed for this particular use case). Both the hvplot and gv-only use cases were run in the same environment (Cartopy was 0.18.0 at the time and GeoViews 1.9.1).