We believe that data-driven decisions can make a difference in a specific moment of someone’s life and our world. Insights require accurate data at the right temporal and spatial resolution. But collecting raw data is not enough to derive insights. Normalizing events related to demography is critical to derive insights and inform business decisions. 

LocalPop is a high-resolution population model derived from nighttime light, the road network, and the Canadian census data. You can explore the model directly from your web browser in Geometric Data Viewer.

We used Carto’s Building Spatial Model to Classify Global Urbanity Levels as inspiration to create the model. However, there are significant differences with our model:

• We use Uber’s H3 grid system with a higher resolution (resolution 8 ~0.75 squared km).
• We provide a numerical estimate of the population for each hexagon rather than an urbanity level.

Some use cases for LocalPop:

• To target which communities are currently underserved and optimized their strategy, a brand must normalize its sales data.
• To evaluate their risk exposure, insurers can use the model to normalize the occurrence of accidents, such as fires or floods.
• To understand urban sprawl, municipalities must have an accurate picture of where their citizens are living

Future Improvements

We built out a data pipeline similarly discussed in our related blog post. Working at the H3 resolution 8 to create derived datasets has some conceptual and technical challenges. The 2016 census data is used to generate the model. We plan to release the 2021 version shortly.

We also aim to use nighttime light to interpolate the growth of the population between the two census releases. We will love to hear from you if you need higher frequency population data. Understanding your use case could be crucial to our design. 

Free Beta Release

We want to learn from the users experimenting with the model in the field. So, if you are interested in being part of our free beta program, reach out to me (zachary@anagraph.io) or Nicolas (nicolas@anagraph.io). 

Other Projects?

We developed our capacity to generate high-resolution models using the best-of-breed open source technologies in geospatial. To scale PostGIS with TB of high-resolution imagery and H3 grids, we got you covered. We can adapt our pipelines with dbt to any cloud provider or infrastructure, small or big data alike.