tomplex / meridian

Geospatial data structure abstractions for Python

howdy

I'm tom. I'm a software engineer primarily focused on working with all kinds of data. I am competent in several programming languages, but Python remains my favorite. I pride myself on being a jack of all trades and as such I have worked on all technical aspects of software projects from design to deploy.

I love to make things regardless of the medium, be it dough, wood, or code.

I currently work at Faraday, where we build developer tools for responsible AI.

meridian's Issues

Create a Docker image with all dependencies installed

There should be a Docker image available with all dependencies installed for quick and hassle-free implementation in that environment.

Add basic geometry operations and __geo_interface__ to `spatialdata` objects

spatialdata objects should implement the __geo_interface__ protocol, and have access to basic geometry operations. Example usage:

import fiona

from meridian import SpatialDataset
from shapely import geometry

with fiona.open('/path/to/my/shapefile.shp') as src:
    dataset = SpatialDataset(src)

poi = geometry.shape({
    'type': 'Point',
    'coordinates': [-72.319261, 43.648956]
})

for record in dataset.intersection(poi):
    print(record)
    record.intersects(poi) # add a few basic geometry methods onto spatialdata objects
    print(record.__geo_interface__) # prints a bunch of JSON representing the feature's geometry and properties

Ideally, we do this without removing namedtuples - I like the simplicity of use. Maybe meridian should take over its implementation of namedtuple so we can add custom methods / properties that are domain-specific.

Create an example directory with some sample data and usage examples

It should go over some of the main functionality and behavior of the SpatialDataset object.

Add documentation using sphinx

Online documentation should be built using Sphinx, and probably hosted on github pages.

Decide if format compatibility should live in meridian, or in another package

There should be helpers available for reading different data formats into meridian, such as PostGIS, WKT, etc. Essentially all they would do is read other files / sources of data and convert each row into a GeoJSON feature.

This need is general enough that it could easily be its own library instead of part of meridian itself.

Include in meridian: pros and cons

pros

• Users don't need to install a second package to get common functionality

cons

• Support for PostGIS (which I think is critical) adds a dependency on psycopg2, which the average user might not care about, and requires a separate binary on Windows.

...will add more as I think of them.

Create comparison of meridian vs GeoPandas

Want to look specifically at how long it takes to load data and perform spatial joins, and perform more advanced processing.