We have invested a bit of time to integrate our traffic visualization tool into R, taking advantage of Shiny and shinyapps.io to deploy an application for visualizing our historically averaged traffic data for each day of the week (see here). We hope this is a fun and interesting tool to play around with. Commuters may even find it useful. (more…)
traffic
Daily traffic evolution and the Super Bowl
With an eye towards predicting traffic evolution, we begin by examining the time-dependence of the contribution from the first principal components on different days of the week. Traffic throughout the day $\vert x(t) \rangle$ can be represented in the basis of principal components; $\vert x(t) \rangle$ $= \sum_{i} c_i(t) \vert \phi_i \rangle $$^1$, where $\vert \phi_i \rangle$ is the ith principle component. The coefficients $c_i(t)$, sometimes called the “scores” of $\vert x(t) \rangle $ in the basis of principal components, carry all of the dynamics.
The largest deviations in the traffic patterns (and of the scores) are during weekday rush hours (around 8 am and 5 pm) – see plot of the scores for several modes throughout Jan. 15. (more…)
Data reduction by PCA
Here, we characterize the data compression benefits of projection onto a subset of the eigenvectors of our traffic system’s covariance matrix. We address this compression from two different perspectives: First, we consider the partial traces of the covariance matrix, and second we present visual comparisons of the actual vs. projected traffic plots. (more…)
Traffic patterns of the year: 2014 edition
As we mentioned in the last post, there are currently over 2000 active speed loop detectors within the Bay Area highway system. The information provided by these loops is often highly redundant because speeds at neighboring sites typically differ little from one another. This observation suggests that a higher level, “macro” picture of traffic conditions could provide more insight: Rather than stating the speed at each detector, we might instead offer info like “101S is rather slow right now”. In fact, we aim to characterize traffic conditions as efficiently as possible. To move towards this goal, we have carried out a principal component analysis (PCA)$^1$ of the full 2014 (year to date) PEMS data set. (more…)
Obtaining and visualizing traffic data
In our first set of posts here, we explore the possibility of using historical traffic data to train a machine learning algorithm capable of predicting near-term highway conditions — say, up to an hour into the future, at any given time. To try our hand at this, we will be working with publicly available data provided by the California Performance Measurement System (PEMS). (more…)