clusterRoutes {openSkies}

Cluster aircraft trajectories based on positional features

Description

Performs clustering of aircraft trajectories positional based on their positional features with several available methods. The input should be either a list of openSkiesStateVectorSet or an already computed features matrix as returned by getVectorSetListFeatures. If the input is a list of vector sets, features will be computed with default settings.

Usage

clusterRoutes(input, method="dbscan", eps=0.5, numberClusters=NULL, ...)

Arguments

input

input to be clustered, given as either a list of openSkiesStateVectorSet, or a matrix of positional features extracted from a list of openSkiesStateVectorSet objects with getVectorSetListFeatures, that will be used to identify clusters.

method

clustering method that will be applied to the positional features. Accepted methods are: dbscan, kmeans, hclust, fanny, clara, agnes

eps

Size of the epsilon neighborhood to be passed to dbscan.This argument is only used if the selected clustering method is dbscan..

numberClusters

number of expected clusters. If NULL or a value lesser than 2 is passed, the number of clusters will be estimated. This argument is only used if the selected clustering method is kmeans, hclust, fanny, clara, or agnes

...

additional arguments accepted by the selected clustering method

.

Value

An object with clustering results, containing at least a "cluster". For additional details, see the documentation of cluster.

Examples

if(interactive()){
# Retrieve series of state vectors for 7 instances of flights between 
# Cagliari-Elmas airport and Parma airport

vectors1=getAircraftStateVectorsSeries(aircraft="4d2219", 
startTime="2020-11-06 09:20:00", endTime="2020-11-06 10:30:00", 
timeZone="Europe/London", timeResolution=300)

vectors2=getAircraftStateVectorsSeries(aircraft="4d226c", 
startTime="2020-10-30 09:20:00", endTime="2020-10-30 10:30:00", 
timeZone="Europe/London", timeResolution=300)

vectors3=getAircraftStateVectorsSeries(aircraft="4d225b", 
startTime="2020-10-29 07:15:00", endTime="2020-10-29 08:25:00", 
timeZone="Europe/London", timeResolution=300)

vectors4=getAircraftStateVectorsSeries(aircraft="4d225b", 
startTime="2020-10-25 06:25:00", endTime="2020-10-25 07:35:00", 
timeZone="Europe/London", timeResolution=300)

vectors5=getAircraftStateVectorsSeries(aircraft="4d224e", 
startTime="2020-10-19 09:30:00", endTime="2020-10-19 10:40:00", 
timeZone="Europe/London", timeResolution=300)

vectors6=getAircraftStateVectorsSeries(aircraft="4d225b", 
startTime="2020-10-16 09:30:00", endTime="2020-10-16 10:30:00", 
timeZone="Europe/London", timeResolution=300)

vectors7=getAircraftStateVectorsSeries(aircraft="4d227d", 
startTime="2020-10-12 09:30:00", endTime="2020-10-12 10:30:00", 
timeZone="Europe/London", timeResolution=300)

# Retrieve state vectors for an outlier flight, corresponding to a flight 
# between the airports of Sevilla and Palma de Mallorca

vectors8=getAircraftStateVectorsSeries(aircraft = "4ca7b3", 
startTime="2020-11-04 10:30:00", endTime="2020-11-04 12:00:00",
timeZone="Europe/London", timeResolution=300)

## Group all the openSkiesStateVectorSet objects in a single list

vectors_list=list(vectors1, vectors2, vectors3, vectors4, vectors5, vectors6, vectors7, vectors8)

## Extract the matrix of features

features_matrix=getVectorSetListFeatures(vectors_list, scale=TRUE, useAngles=FALSE)

## Perform clustering

clustering=clusterRoutes(features_matrix, "dbscan", eps=5)

## Display clustering results with flights colored by assigned cluster

plotRoutes(vectors_list, pathColors=clustering$cluster, literalColors=FALSE)
}