cuSpatial Python User Guide #666
Conversation
More writing. Geopandas and cudf integration Formatting it. Updating with better code examples. Adding examples.
|
This PR is nearly complete with an incoming |
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
This looks all very promising and nice to read, I just left some comments mostly about the introduction section on the terms and phrases that I got confused about. The code section is mostly good. |
There was a problem hiding this comment.
This is a great. I have a number of edits / suggestions. GitHub started doing weird things with my suggestions towards the end (treating the original text as an encoded image).
My biggest concern is that this document is already getting quite long, and as cuSpatial's API grows it is going to get unwieldy. I think going forward we should explore how we can modularize it, and where possible, associate the interactive examples with the individual API doc sections/pages.
|
I've addressed the existing reviews. |
| "## GPU accelerated memory layout\n", | ||
| "\n", | ||
| "cuspatial uses `GeoArrow` buffers, a GPU-friendly data format for geometric data that is well \n", | ||
| "suited for massively parallel programming. See [I/O](#io) on the fastest methods to get your \n", |
There was a problem hiding this comment.
Test that this link works.
| "suited for massively parallel programming. See [I/O](#io) on the fastest methods to get your \n", | |
| "suited for massively parallel programming. See [I/O](#input-output) on the fastest methods to get your \n", |
| @@ -0,0 +1,1602 @@ | |||
| { | |||
There was a problem hiding this comment.
a Tuple of GeoArrow buffers
But... they are just plain cudf series and dataframes!
I would propose we implement `GeoSeries.from_filewith this function.
Reply via ReviewNB
There was a problem hiding this comment.
I think I'd like to update read_polygon_shapefile to return a GeoSeries, as from_file implies many more file types to support. We could write from_file and check if it is a shapefile, then use our internal reader otherwise use GeoPandas, but performance would be poor for all other file types. What do you think? #715
|
@thomcom I propose removing from this PR the speculative sections at the end on building up the rest of DE9IM (contains, etc) using existing primitives. That could be saved for a followup PR. |
|
I'll remove the stuff at the bottom and then merge. |
There was a problem hiding this comment.
I have these comments still open:
#666 (comment)
#666 (comment)
#666 (comment)
|
@gpucibot merge |
This PR adds
docs/source/user_guide/index.ipynb, a comprehensive description of the APIs available in cuspatial python and example code for each.The APIs are separated in a table of contents by the logical structure we've settled on for their location in the API. The demonstrate uses the
cuspatial.version of every API, none are accessed using their full python package address.At the bottom I'm working on some examples of performing
containsandwithin, DE-9IM operations that can be performed easily withpoint-in-polygon. Otherwise the document is complete.Depends on #685 and #680
This contributes to or closes #599 by providing an example use case of all python APIs.