Introduction
Today’s public transit vehicles produce a large amount of automatic
vehicle location (AVL) data. This data is very useful for planning and
performance studies, but can be noisy, error-prone, and sparse. We
present transittraj, an R package which cleans AVL point
data and reconstructs continuous, differentiable, monotonic, and
invertible vehicle trajectory functions.
These vignettes provide code to support the transittraj
paper (O’Brien and Lehe,
2026), including all data cleaning, analysis, and visualization
code. Unfortunately, we do not have permission to share the underlying
data, though we encourage the reader to replicate our analyses with
their own data. The vignettes presented here are not intended as
stand-alone articles; for a beginner-friendly introduction to
transittraj with open sample data, check out our tutorial
vignettes (vignette("articles/input-data-la")).
Navigating these Vignettes
We’ve split the code used in the paper into three vignettes:
The AVL Cleaning Workflow. This vignette includes the code for the visualizations presented in Section 2 (Package Architecture and Workflow) and the cleaning results presented in Section 3 (Case Study).
Estimating Signal Performance Measures Using Reconstructed Trajectories. This vignette includes all code used to estimate and visualize the traffic signal performance measures presented in Section 3 (Case Study).
Evaluating the Robustness of Reconstructed Trajectories. This vignette includes all code used to quantify and visualize the error in reconstructed trajectories across various polling frequencies and interpolating methodologies, as discussed in Section 4 (Robustness Check).
Thorough discussions of the results are presented in our paper. In
addition to the knit vignettes, the raw Rmarkdown files are
available at our supplementary
GitHub repository. These include all raw code chunks, including
those hidden in the knit vignettes.