DRHotNet: An R package for detecting differential risk hotspots on a linear network

TL;DR

DRHotNet is an R package designed to detect differential risk hotspots along linear networks, such as roads, providing more precise insights into localized event concentrations and overrepresentations compared to traditional areal analysis.

Contribution

This paper introduces DRHotNet, an R package that enables detection of differential risk hotspots on linear networks, addressing a gap in existing spatial analysis tools.

Findings

Successfully applied to crime data for hotspot detection

Facilitates analysis at the road segment level

Enhances understanding of event overrepresentation

Abstract

One of the most common applications of spatial data analysis is detecting zones, at a certain investigation level, where a point-referenced event under study is especially concentrated. The detection of this kind of zones, which are usually referred to as hotspots, is essential in certain fields such as criminology, epidemiology or traffic safety. Traditionally, hotspot detection procedures have been developed over areal units of analysis. Although working at this spatial scale can be suitable enough for many research or practical purposes, detecting hotspots at a more accurate level (for instance, at the road segment level) may be more convenient sometimes. Furthermore, it is typical that hotspot detection procedures are entirely focused on the determination of zones where an event is (overall) highly concentrated. It is less common, by far, that such procedures prioritize the location of zones where a specific type of event is overrepresented in relation to the other types observed, which have been denoted as differential risk hotspots. The R package DRHotNet provides several functionalities to facilitate the detection of differential risk hotspots along a linear network. In this paper, DRHotNet is depicted and its usage in the R console is shown through a detailed analysis of a crime dataset.