George Willis / Data scientist
High-resolution nocturnal satellite radiometry nighttime lights map of China and East Asia
Data scientist · London

Open data and a map or two.

I’m George, a data scientist at Cisco in London. I work with messy network data, investigate unusual patterns and work out what the measurements can actually tell us.

Outside work, I build things around music, transport and cities. They usually start with a question or a small annoyance, then become an excuse to learn something new. The scope is not always well contained.

· China’s night lights, 2000–2019 → · DMSP-OLS & VIIRS · Chen et al. (2021)
Jing-Jin-Ji Megalopolis
39.90° N, 116.40° E · Beijing–Tianjin–Hebei
Sichuan Basin
30.57° N, 104.06° E · Chengdu–Chongqing
Yangtze River Delta
31.23° N, 121.47° E · Shanghai–Nanjing
Pearl River Delta
23.13° N, 113.26° E · Guangzhou–Shenzhen
NIGHTTIME LIGHTS INTENSITY (LOG SCALE)
Low
High
SOURCE: CHEN ET AL. (2021) · DMSP-OLS & VIIRS (2000–2019) · R / RASTER

Selected work

All work →
GIS · Network Simulation

TransitFlow: metro network modelling

A fun experiment in building a metro-network model, starting with Beijing and seeing whether the same approach could work across other cities.

Python · NetworkX · Leaflet · pyproj TransitFlow Beijing metro network model preview thumbnail
GIS · Spatial Regression

House prices and London rail access

Initially built to help my sister locate affordable commuter towns. How much does access to the railway matter for London house prices, and does the relationship look the same everywhere?

R · Spatial Modeling · Leaflet House prices and London rail access preview thumbnail
Machine Learning · Network Regression

Regional hyperlinks and trade flows

Some of the method behind the publication. Can links between websites tell us something about trade between regions? Research combining web data, network analysis and predictive modelling. Published in the Annals of the American Association of Geographers.

R · Python · Network Analysis · Applied Statistics Regional hyperlinks preview thumbnail
GIS · Travel Catchments

Location choice: London travel catchments

Exploring which parts of London are within reach of different stations, so i could decide where to live with my mates.

R · Isochrone Modeling · Leaflet London locations preview thumbnail
GIS · Spatial Kernel Density

Finding Beijing’s urban extent

Combining satellite night lights and points of interest to investigate where Beijing’s built-up area ends and its surroundings begin.

Python · Spatial Analysis · Remote Sensing Beijing urban extent preview thumbnail
GIS · Satellite Radiometry

China’s night lights, 2000–2019

Twenty years of harmonised satellite night-light data, brought together in an animation to explore how patterns of light changed across China.

Python · R · Raster Analysis · Time Series China animated nightlights preview thumbnail

Writing & research

All articles →
Build notes TransitFlow

Modelling the Beijing metro

Could I build a metro-network model and then reuse it somewhere else? The data, services and network ideas behind TransitFlow, drawing on my earlier research in spatial networks.

Published · 2021 Travel Behaviour and Society

Uber and New York’s yellow taxis

My undergraduate dissertation examined the relationship between Uber’s growth and yellow taxi demand in New York. It later became a paper in Travel Behaviour and Society.

Publications

All publications →
2022

Using the web to predict regional trade flows

Research into whether hyperlinks between websites can help predict trade between regions. We combined web-network data with economic data and tested how well the models worked beyond the period used to train them. With Emmanouil Tranos and André Carrascal-Incera.

Tranos, E., Carrascal-Incera, A. & Willis, G. · Annals of the American Association of Geographers, 2022
@article{tranos2022using,
  title={Using the web to predict regional trade flows: data extraction, modelling, and validation},
  author={Tranos, Emmanouil and Carrascal-Incera, Andr{\'e} and Willis, George},
  journal={Annals of the American Association of Geographers},
  volume={113},
  number={1},
  pages={168--189},
  year={2022},
  publisher={Taylor \& Francis},
  doi={10.1080/24694452.2022.2109577}
}
2021

Using "Big Data" to understand the impacts of Uber on taxis in New York City

Research examining Uber’s growth alongside changes in yellow taxi ridership, combining New York taxi records, contextual data and time-series analysis. With Emmanouil Tranos.

Willis, G. & Tranos, E. · Travel Behaviour and Society, 22: 94-107, 2021
@article{willis2021using,
  title={Using "Big Data" to understand the impacts of Uber on taxis in New York City},
  author={Willis, George and Tranos, Emmanouil},
  journal={Travel Behaviour and Society},
  volume={22},
  pages={94--107},
  year={2021},
  publisher={Elsevier},
  doi={10.1016/j.tbs.2020.08.003}
}