logs/week15.md

In wjones127/thesis: Bike Routes Thesis (Title Case)

Week 15

Possible models

Here I'd like to write down several possible models. One new idea I've had for modelling is using segment popularity on the segment level as a predictor for ride rating.

rating ~ (1 | rider) + temp + rush.hour +

Map Matching

I keep on coming back to this. More and more I get the sense that the map matching they have done won't be sufficient. It looks like the routes have nearly been roughly snapped, but a lot of strays, rather than actually mapped to segments. I may have to quickly find a way to snap it.

This could be really good though, if I am able to match it directly to the bicycle network from CivicApps.

Here is one promising library: https://github.com/Project-OSRM/osrm-backend/wiki

Progress on cleaning data

I've finally started actually cleaning most of my data better. I've created the folder data-raw to hold the raw data as well as the scripts that clean the data. The nice thing here is that I can now just have those scripts run on the data when I run the models.

wjones127/thesis documentation built on May 4, 2019, 7:34 a.m.