# St Andrews meeting: Popgetter and router demo

## Goals for the demo

- Show off the existing work on popgetter and the router
- Show how integrated tools can speed up our ability to deliver projects
- Show how these tools can allow us to port an existing project to a new geography with little work .

## Outline of the presentation

- Start with the description on an example problem: A national project to locate potential problem intersections / roads along likley walk to school paths.
- Break down the problem into steps
1. Determine where the population of school-going children live
2. Identify the schools they go to
3. Identify the most likely routes they take to take to get there.
4. Find the most unsuitable roads/intersections used by most children.

- Hi-light that these steps, determining population distribution, identifying resources and finding paths between them are problems we encounter repeatedly on projects. We want to find way to make this faster and easier

### Popgetter demo

- Show a brief diagram of the popgetter DAG and explain about the ETL process a little
- Use the web tool to see what populations are available for a given region and identify the population of interest we have
- Extract that population using the python lib (backed by rust) or command line tool to a file or dataframe

### Router demo

- Start with a diagram of how the routing engine works. (What pieces are configurable / required by the user to input, how the pipeline works, what the output shows)
- Now that we have the population starting locations and the schools (we can say that this was provided by our partner) we need to find routes to them.
- Quick identification of OD pairs by simple nearest crow fly distance between geometry centroid and school
- Running the router between these to get the most likely routes
- Adding up counts along road segments to identify the most used.

### A new request from another country

- Explain that after publication of these results we had a request from a state in America to do the same thing.
- Show that popgetter + the router allow this analysis to happen by just changing a few inputs rather than having to develop a new pipeline
- Quickly show the results

### Discussion of future work and challenges

For popgetter
- Harmonization of data from different countries
- Other ways of consuming the data
- Potentially moving beyond population datasets

For the router
- The ability to quickly and easily tweak edge weights to represent different populations transit preferences
- Building on the router to support different access models
- The challenge of defining and validating edge cost functions and Level of Traffic Stress definitions against real behaviour