Kernel Density Estimation on a Linear Network

© 2016 Board of the Foundation of the Scandinavian Journal of Statistics.This paper develops a statistically principled approach to kernel density estimation on a network of lines, such as a road network. Existing heuristic techniques are reviewed, and their weaknesses are identified. The correct an...

Full description

Bibliographic Details
Main Authors: Mcswiggan, G., Baddeley, Adrian, Nair, G.
Format: Journal Article
Published: Blackwell Publishing Ltd 2016
Online Access:http://hdl.handle.net/20.500.11937/52531
_version_ 1848758949559402496
author Mcswiggan, G.
Baddeley, Adrian
Nair, G.
author_facet Mcswiggan, G.
Baddeley, Adrian
Nair, G.
author_sort Mcswiggan, G.
building Curtin Institutional Repository
collection Online Access
description © 2016 Board of the Foundation of the Scandinavian Journal of Statistics.This paper develops a statistically principled approach to kernel density estimation on a network of lines, such as a road network. Existing heuristic techniques are reviewed, and their weaknesses are identified. The correct analogue of the Gaussian kernel is the 'heat kernel', the occupation density of Brownian motion on the network. The corresponding kernel estimator satisfies the classical time-dependent heat equation on the network. This 'diffusion estimator' has good statistical properties that follow from the heat equation. It is mathematically similar to an existing heuristic technique, in that both can be expressed as sums over paths in the network. However, the diffusion estimate is an infinite sum, which cannot be evaluated using existing algorithms. Instead, the diffusion estimate can be computed rapidly by numerically solving the time-dependent heat equation on the network. This also enables bandwidth selection using cross-validation. The diffusion estimate with automatically selected bandwidth is demonstrated on road accident data.
first_indexed 2025-11-14T09:52:07Z
format Journal Article
id curtin-20.500.11937-52531
institution Curtin University Malaysia
institution_category Local University
last_indexed 2025-11-14T09:52:07Z
publishDate 2016
publisher Blackwell Publishing Ltd
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-525312017-09-13T15:39:24Z Kernel Density Estimation on a Linear Network Mcswiggan, G. Baddeley, Adrian Nair, G. © 2016 Board of the Foundation of the Scandinavian Journal of Statistics.This paper develops a statistically principled approach to kernel density estimation on a network of lines, such as a road network. Existing heuristic techniques are reviewed, and their weaknesses are identified. The correct analogue of the Gaussian kernel is the 'heat kernel', the occupation density of Brownian motion on the network. The corresponding kernel estimator satisfies the classical time-dependent heat equation on the network. This 'diffusion estimator' has good statistical properties that follow from the heat equation. It is mathematically similar to an existing heuristic technique, in that both can be expressed as sums over paths in the network. However, the diffusion estimate is an infinite sum, which cannot be evaluated using existing algorithms. Instead, the diffusion estimate can be computed rapidly by numerically solving the time-dependent heat equation on the network. This also enables bandwidth selection using cross-validation. The diffusion estimate with automatically selected bandwidth is demonstrated on road accident data. 2016 Journal Article http://hdl.handle.net/20.500.11937/52531 10.1111/sjos.12255 Blackwell Publishing Ltd restricted
spellingShingle Mcswiggan, G.
Baddeley, Adrian
Nair, G.
Kernel Density Estimation on a Linear Network
title Kernel Density Estimation on a Linear Network
title_full Kernel Density Estimation on a Linear Network
title_fullStr Kernel Density Estimation on a Linear Network
title_full_unstemmed Kernel Density Estimation on a Linear Network
title_short Kernel Density Estimation on a Linear Network
title_sort kernel density estimation on a linear network
url http://hdl.handle.net/20.500.11937/52531