This is an attempt to visualize the results of some ongoing work to represent traffic patterns in Manhattan via low-dimensional 'signatures'. The data stems from taxi data from 2011 and builds on the work of Brian Donovan and Daniel Work “Using coarse GPS data to quantify city-scale transportation system resilience to extreme events” (Transportation Research Board 94th Annual Meeting, January 2015).
In our current work, we (roughly) start with a big matrix consisting of traffic counts organized by roads (columns) and times (rows). We then carry out a low-rank factorization (i.e., collaborative filtering) which gives us robust spatial signatures which are modulated by time.
The above a map (which may take a long time to load) which shows how different roads are decomposed into different patterns. For example, West 38th Street between 5th and 6th Avenues is about 55% pattern 3 and 45% pattern 39.
The above shows how different spatial signatures are modulated by time over the course of week.