Public Transport Analyser

This software is designed to see how much longer it takes to use public transport compared to driving.

The algorithm

The basic idea is that the program generates a random start location (the origin), and then 10 random end locations (the destination(s)). For each route (from origin to destination), it asks Google for how long it will take to drive, and then how long it will take using public transport at various times (currently: 6:00, 8:00, 12:00, 17:00, 21:00).

The ratio of time from driving to public transport is then calculated to see how good or bad it is. After these 10 random destinations are done, a new origin is generated, and the process repeats. After some time, the algorithm will revisit previous origins, and add more destinations to provide better coverage. It runs at about 1000 m resolution at the moment, but will be increased.

The idea

Hopefully the end product will be a website where someone can click on a point on a map, and it will display a heat map of the surrounding city, with the colours representing how good or bad the public transport system can service the route (compared to owning a car). This will hopefully be a tool the public and government can use to assess where new public transport upgrades should be targeted. It also may reveal certain biases, like if safe political seats get better or worse treatment in these sort of public services. This idea was partly inspired by this.

Current issues

To use Google's distance matrix API for free, it is limited to 2500 requests a day. Currently each route takes 6 requests (1 driving, 5 public transport times), and each origin gets 10 routes generated. This means about 41 routes per day, which is pretty slow going. Oh well.

UPDATE: I'm now on the free 60 day trial, so it's much faster now.

Can I run this too?

Yes, you just need to make two files:

1. public_transport_analyser/data_gatherer/api.py

and in there put your Google Maps Distance Matrix API key. It should look like:

apikey="AJAH<your actual key>SJKAJ"

2. public_transport_analyser/database/db_details.py

and it should look like:

dbusername = "<your db username>"
dbpassword = "<your db password>"
dbname = "<the name of your db>"

Dependencies

For the server aspect (main.py), you need:

• arrow 0.7.0 0.7.0
• pony 0.6.5 0.6.5
• requests 2.9.1 2.9.1
• plus whatever they depend on

For the visualiser (visualise_routes.py), you need:

• pony 0.6.5 0.6.5
• matplotlib
• plus whatever they depend on

The visualiser is pretty dodgey at the moment, just used for debugging.

Code analysis

Things to do

• Batch requests to Google. Currently it's 1 request for 1 route, but they support up to 100. Need to balance the time though.
• Fix up visualiser.
• Think about how to do web vis.