Welcome to UrbanPy :city_sunrise:

A library to download, process and visualize high resolution urban data in an easy and fast way.

UrbanPy is an open source project to automate data extraction, measurement, and visualization of urban accessibility metrics.

Functional goals

• Download open source spatial data (Limits & Points of Interests)
• Allow for the use of a grid system or administrative boundaries as spatial units.
• Origin-destination matrix calculation by any mode using a routing API
• Obtain travel time from spatial units to the closest facilities
• Consolidate the results as tables and/or shapefiles (georeferenced datasets)
• Visualise the results as maps

UX goals

• Atomic functions (one purpose per function)
• Use the power of Python Geospatial Ecosystem under the hood
• Allow to flexible processing pipelines (custom layer/metrics aggregations)
• Clear documentation with usage and examples
• Clear and replicable example notebooks

Main modules

• download: Main functions for data download from Nominatin API, OverPass API and HDX population data
• geom: Spatial operations, grid partitioning, spatial filtering and street network statistics
• plotting: Visualization wrappers for plotly interactive choropleth maps
• routing: Distance matrix computations (may require your own API keys)
• utils: Data handling helpers

Installation

For users

To install the urbanpy library you can use:

$ pip install urbanpy

Then use import urbanpy in your python scripts to use the library.

If you plan to use the OSRM Server route or distance matrix calculation functionalities* you must have Docker installed in your system, refer to Docker Installation.

Additional Dependecies Notes

• It is important to note that for travel time computation, if needed, a method is implements the Open Source Routing Machine (OSRM). This method pulls, extracts and adds graph weights to the downloaded network and runs the routing server. Make sure to have docker installed for the library to work correctly.

• Urbanpy provides a simple approximation with nearest neighbor search using a BallTree and haversine distance, but the difference between real travel time and the approximation may vary from city to city.

• Additionally, the use of spatial libraries like osmnx, geopandas and h3 require certain extra packages. Specifically, for rtree (spatial indexing to allow spatial joins) libspatialindex is required. OSMnx and Geopandas requiere GDAL as well. If not handled by installing geopandas's dependencies, installing fiona, pyproj and shapely should satisfy the requirements. Another way to ensure all dependencies are met, installing osmnx via conda should suffice. H3 requires cc, make, and cmake in your $PATH when installing, otherwise installation will not be successful. Please refer to h3's documentation for a more detailed guide on installation options and requirements.

Examples

UrbanPy lets you download and visualize city boundaries extremely easy:

import urbanpy as up

boundaries = up.download.nominatim_osm('Lima, Peru', expected_position=2)
boundaries.plot()

Since boundaries is a GeoDataFrame it can be easily plotted with the method .plot(). You can also generate hexagons to fill the city boundaries in a oneliner.

hexs, hexs_centroids = up.geom.gen_hexagons(resolution=9, city=boundaries)

Also check our example notebooks, and if you have examples or visualizations of your own, we encourage you to share contribute.

For developers

If you plan to contribute or customize urbanpy first clone this repo and cd into it. Then, we strongly recommend you to create a virtual environment. You can use conda, this installation manage some complicated C spatial library dependencies:

$ conda env create -n urbanpy -f environment.yml python=3.6
$ conda activate urbanpy

Or if you are more confident about your setup, you can use pip:

$ python3 -m venv .env
$ source .env/bin/activate
(.env) $ pip install -r requirements.txt

License

Copyright © 2020. Banco Interamericano de Desarrollo ("BID"). Uso autorizado. AM-331-A3

Authors

UrbanPy's original authors are Claudio Ortega (socials), Andrés Regal (socials), and Antonio Vazquez Brust (socials).

Contribution guidelines

If you want to contribute to UrbanPy, be sure to review the contribution guidelines. This project adheres to UrbanPy's code of conduct. By participating, you are expected to uphold this code.

*Current support is tested on Linux Ubuntu 18.04 & Mac OS Catalina, coming soon we will test and support Windows 10.