dayanaa
commented
7 years ago Still looking for some more examples of the geostatistical method below in case it's unclear here, but here is where the content stands as of the early afternoon.
Pull populations at census block level through API and calculate it against buffers
- Find population by census block (if unavailable, the next highest geographic level is census block group)
- Clip buffer on census blocks
- Measure what percentage of area is within clipped buffer (so for example, you should then theoretically be able to say that 20% of census block A is within Buffer 1)
- Take the percentage of how much of a census block is within the buffer and multiply that by the number of population within that census block (so for example: if 1000 people live within block A, and 20% of block A is within the buffer, then 20% of 1000 is the spatially calculated number of people living within buffer 1 as it intersects with block A).
- Sum all of the clipped calculations to get the estimated number of people within a buffer
This process is replicable for any variables reported at the census geographic levels - race, income, gender, etc.
Data services, metadata and information:
Eventually, WABA wants to analyze how many people do not live within 1 mile of a bike lane. They also want to see how many more people would live closer to a bike lane if they advocated for new bike lanes in underserved areas.
This will involve pulling in population data from either an API (preferable) or from a dataset. This is a complicated problem that will take us a while to solve, but we can get started on researching this to evaluate the lift and feasibility.
@dayanaa - can you add your research notes and pseudocode from today's Tech Lady hackathon to this ticket?