Research Article

Classifying Vehicle Activity to Improve Point of Interest Extraction

Table 1

Notations used in this paper.

NotationDescription

The set of trajectories
The trajectory within , a strictly ordered sequence of instances
The set of vehicle signals
The values of the vehicle signals at time
An instance is a latitude and longitude position, , and the values of vehicle signals at time
A matrix of the real-valued signals in for the time interval from to , i.e.,
A matrix of the categorical (nominal) signals in for the time interval from to
A matrix of the binary signals in for the time interval from to
A strictly ordered subsequence of instances in trajectory
The set of clusters extracted from all trajectories in
The cluster of trajectory , defined as a strictly ordered subsequence of instances , where is the first instance and is the last instance temporally
A function that returns the time of the instance
A function that returns the first instance in cluster
A function that returns the last instance in cluster
A function that deletes the cluster
A function that returns an array of training and validation clusters for a given number of folds,
A function that returns the ground truth classification label for
A function that returns the AUC
A function that returns the feature values in for the features that are present in feature set
The operator is used to denote the concatenation of two sequences
The set of features that can be extracted from
The set of possible classification labels
The set of positive classification labels,
A pretrained classifier
The feature set used in the classifier
A prediction from the classifier