The Human Mobility Index Python package - hmi

The hmi package is a python module that provides an interface to compute distances between locations based on Özak's (2018, 2010) Human Mobility Index. There are three main classes of computations based on the Human Mobility Index:

• HMI: Only land based mobility
• HMISea: Land based mobility and pre-1500CE seafaring technologies
• HMIOcean: Land based mobility and post-1500CE seafaring technologies

It includes tools to

• Compute distances between locations
• Compute isolation measures for locations (see e.g. Ashraf, Galor, and Özak (2010))
• Generate optimal routes as shape files
• Export optimal travel times, routes, etc. to shape files, CSV files, etc

Install

The package depends on GDAL (through georasters and rasterio), which pip cannot build without a system GDAL library. The best way to install is to first create a mamba/conda environment with the geospatial dependencies from conda-forge:

mamba create --name HMI --override-channels -c conda-forge python=3.12 pip gdal geopandas georasters rasterio scikit-image requests tqdm

activate the environment

mamba activate HMI

then

pip install hmi

or

pip install git+https://github.com/human-mobility-index/hmi.git

The first time you import hmi, the package downloads the HMI cost surfaces (about 4.5 GB) from Zenodo into its data/ folder. This is done only once.

Example Usage: HMI

Sources and destinations are (Geo)Pandas data frames with latitude and longitude (or projected CEA x/y) columns. Distances are measured in weeks of travel.

import pandas as pd
import hmi

sources = pd.DataFrame({'name': ['Addis Ababa', 'London'],
                        'lat': [9.03, 51.51], 'lon': [38.74, -0.13]})
destinations = pd.DataFrame({'name': ['Cairo', 'Beijing', 'Sydney'],
                             'lat': [30.04, 39.90, -33.87], 'lon': [31.24, 116.41, 151.21]})

# Land-only mobility (use hmi.HMISea to add pre-1500CE seafaring)
A = hmi.HMI(sources, destinations, lat='lat', lon='lon')
A.correct_data()             # move points that fall off the cost surface onto it
A.HMIdistance(routes=True)   # travel times, isolation and optimal routes
A.hmidist                    # GeoDataFrame with one row per reachable pair

# Regions closest (in travel time) to each source
regions = A.voronoi()

Pairs that cannot be reached (e.g. London over land only) are dropped from hmidist. Use A.HMIdistance(export_shape=True, export_raster=True, path='./') to write the routes as GeoJSON and the travel-time surfaces as GeoTIFFs.

Testing

From a clone of the repository, inside the environment above:

pip install pytest
pytest

The tests compute HMI and HMISea distances, isolation, routes and Voronoi regions for a few world cities and compare them with known values.

Citation

If you use the package please cite:

Özak, Ömer. "Distance to the pre-industrial technological frontier and economic development." Journal of Economic Growth 23.2 (2018): 175-221.

Özak, Ömer. "The voyage of homo-economicus: Some economic measures of distance." Department of Economics, Brown University (2010).

and the package itself

Özak, Ömer. “The Human Mobility Index Python Package - hmi”. Zenodo, December 7, 2024. https://doi.org/10.5281/zenodo.14291912.

Issues

Find a bug? Report it via github issues by providing

• a link to download the smallest possible raster and vector dataset necessary to reproduce the error
• python code or command to reproduce the error
• information on your environment: versions of python, gdal and numpy and system memory

Copyright

© Ömer Özak (2014)

This code and data is provided under Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) License and GPLv3.