dbscan_te: Determine stop locations in trajectory data via DBSCAN-TE

In byu-transpolab/gpsactivs: Extract Activities and Semantic Locations from GPS Trace Data

Description

Determine stop locations in trajectory data via DBSCAN-TE

Usage

dbscan_te(trajectory, eps = 25, minpts = 4, delta_t = 300, entr_t = 1.75)

Arguments

trajectory	A simple features collection of points representing a single trajectory (i.e. person, day). MUST have a 'dttm' column called "timestamp" representing the time of GPS point.
eps	distance threshold for defining clusters. Should be provided in the same units as the trajectory projection (usually meters). DBSCAN input.
minpts	minimum number of points in a cluster. DBSCAN input.
delta_t	time threshold (seconds): a gap of this length within a spatial cluster will split the cluster into two potential activities. An activity must also be at least this long.
entr_t	entropy threshold: the entropy of a cluster is a function of the chaotic movement between points in a cluster. Clusters with higher entropy are more likely to be activities; this parameter will exclude potential activities below this threshold

Details

Implements the method described in Gong, L., Yamamoto, T., & Morikawa, T. (2018). Identification of activity stop locations in GPS trajectories by DBSCAN-TE method combined with support vector machines. Transportation Research Procedia, 32, 146–154. https://doi.org/10.1016/J.TRPRO.2018.10.028

Note: the SVM is not (yet) implemented. The entropy calculation and heuristics seem to get us almost all the way there.

Value

A simple features collection of points with the start and end time of each activity, the estimated cluster entropy (for debugging / calibrating)

byu-transpolab/gpsactivs documentation built on Dec. 19, 2021, 12:51 p.m.