kth.sePublications KTH
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Random forest modelling of remotely sensed land cover data to identify crime hot spots in urban areas
KTH, School of Architecture and the Built Environment (ABE), Urban Planning and Environment, Urban and Regional Studies.ORCID iD: 0000-0002-0529-4824
KTH, School of Architecture and the Built Environment (ABE), Urban Planning and Environment, Urban and Regional Studies.ORCID iD: 0000-0001-9692-8636
KTH, School of Architecture and the Built Environment (ABE), Urban Planning and Environment, Urban and Regional Studies.ORCID iD: 0000-0001-5302-1698
Department of Geography, University of Cambridge, Cambridge, UK.ORCID iD: 0000-0003-3462-7218
2025 (English)In: Discover Cities, E-ISSN 3004-8311, Vol. 2, no 122Article in journal (Refereed) Published
Abstract [en]

This study evaluates the effectiveness of integrating high-resolution remote sensing (RS) data with machine learning (ML) techniques to identify criminogenic environments in urban areas. We employ an unsupervised ISO clustering method to classify land cover from aerial imagery, thereby capturing fine-scale environmental details that are often overlooked in traditional analyses. These clusters are linked to both crime and non-crime events through a presence/absence (case–control) framework, a methodology adapted from species distribution studies, which enables a micro-environmental examination of crime locations. In addition to RS-derived land-cover predictors, the study incorporates socio-economic and demographic variables, as well as a centrality indicator that proxies the intensity of urban activity. A Random Forest classifier is utilized to model the likelihood of street theft incidents based on these predictors. The model achieves robust performance, with an F1-score of 0.88 ± 0.03 as determined by K-fold cross-validation. To enhance model interpretability, SHapley Additive exPlanations (SHAP) is applied. The findings of this research demonstrate that integrating RS data with ML techniques offers a valuable tool for identifying and mapping criminogenic environments. The resulting risk map of Stockholm highlights key urban areas with elevated street theft risk, offering guidance for targeted crime prevention and urban planning strategies. While our workflow simplifies some technical steps compared to other RS + ML pipelines, it still requires GIS and ML competence to implement effectively. This approach reduces, but does not eliminate, sensitivity to spatial unit choice (MAUP) and spatial data dependencies.

Place, publisher, year, edition, pages
Springer Nature, 2025. Vol. 2, no 122
Keywords [en]
Random forest classifier, Remote sensing, Street theft, Prediction, Cross-validation, Presence/absence
National Category
Geosciences, Multidisciplinary
Research subject
Planning and Decision Analysis, Urban and Regional Studies
Identifiers
URN: urn:nbn:se:kth:diva-354128DOI: 10.1007/s44327-025-00171-2OAI: oai:DiVA.org:kth-354128DiVA, id: diva2:1901716
Funder
Swedish Research Council Formas, 2020-01999
Note

QC 20251214

Available from: 2024-09-30 Created: 2024-09-30 Last updated: 2026-01-19Bibliographically approved
In thesis
1. Understanding crime patterns using spatial data analysis: Case studies in Stockholm, Sweden
Open this publication in new window or tab >>Understanding crime patterns using spatial data analysis: Case studies in Stockholm, Sweden
2024 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

Understanding the complex relationship between urban environments and crime is crucial for effective urban planning and crime prevention strategies. Spatial analytical methods have provided valuable knowledge into crime patterns, enabling the detection of crime-concentrated environments and informing law enforcement operations and urban planning interventions. The international literature highlights the increasing use of remote sensing in crime analysis, driven by improved data availability and accuracy. Given the potential of this approach, this thesis investigates the use of spatio-temporal data analyses, particularly the incorporation of remote sensing data along with traditional socio-demographic and land use indicators in understanding the dynamics of crime in urban environments. Four crime categories—cannabis-related crimes, street theft, residential burglaries, and sexual crimes—are investigated using Stockholm City in Sweden as a case study. Remote sensing data, particularly very high-resolution imagery, combined with machine learning algorithms, such as the Random Forest classifier, facilitate the prediction of crime risk areas and the identification of environmental factors associated with crime occurrences. While the thesis reflects upon the advantages and disadvantages of using remote sensing in crime analyses, findings offer practical insights for policymakers, urban planners, and law enforcement agencies, enabling the development of data-informed strategies to foster safer and more resilient urban environments.

Abstract [sv]

Förståelsen av det komplexa sambandet mellan urbana miljöer och brott är avgörande för effektiv stadsplanering och brottsförebyggande strategier. Rumsliga analytiska metoder har gett värdefulla insikter om brottsmönster, vilket möjliggör identifiering av miljöer med hög brottskoncentration och ger underlag för brottsbekämpningsinsatser och stadsplaneringsåtgärder. Den internationella litteraturen lyfter fram den ökande användningen av fjärranalys inom brottsanalys, vilket drivs av förbättrad dataåtkomst och noggrannhet. Med tanke på potentialen i denna metod utforskar denna avhandling användningen av spatio-temporala dataanalyser, särskilt införlivandet av fjärranalysdata tillsammans med konventionella socio-demografiska och markanvändningsindikatorer för att förstå dynamiken kring brott i urbana miljöer. Fyra brottskategorier – cannabisrelaterade brott, fickstölder, bostadsinbrott och sexualbrott – undersöks med Stockholms stad i Sverige som fallstudie. Fjärranalysdata, särskilt bilder med mycket hög upplösning, i kombination med machine learning-algoritmer som Random Forest-klassificeraren, underlättar prediktionen av brottsriskområden och identifieringen av miljöfaktorer som är förknippade med brottsförekomster. Även om avhandlingen reflekterar över fördelar och nackdelar med att använda fjärranalys i brottsanalys, erbjuder våra resultat praktiska insikter för beslutsfattare, stadsplanerare och brottsbekämpande myndigheter, vilket möjliggör utvecklingen av datainformerade strategier för att främja säkrare och mer motståndskraftiga urbana miljöer.

Place, publisher, year, edition, pages
Stockholm: KTH Royal Institute of Technology, 2024. p. 50
Series
TRITA-ABE-DLT ; 2423
Keywords
Exploratory spatial data analysis, random forest classifier, regression analysis, remote sensing data, spatial crime analysis, Utforskande rumslig dataanalys, random forest-klassificerare, regressionsanalys, fjärranalysdata, rumslig brottsanalys
National Category
Geosciences, Multidisciplinary
Research subject
Planning and Decision Analysis, Urban and Regional Studies
Identifiers
urn:nbn:se:kth:diva-354130 (URN)978-91-8106-075-1 (ISBN)
Public defence
2024-10-25, Kollegiesalen, Brinellvägen 8, KTH Campus, https://kth-se.zoom.us/s/66904913390, Stockholm, 09:00 (English)
Opponent
Supervisors
Projects
Development of remote sensing data use for safe environment planning
Funder
Swedish Research Council Formas, 2020-01999
Note

QC 241002

Available from: 2024-10-02 Created: 2024-10-01 Last updated: 2024-10-03Bibliographically approved

Open Access in DiVA

fulltext(4879 kB)32 downloads
File information
File name FULLTEXT01.pdfFile size 4879 kBChecksum SHA-512
2485b65bb122dccf35cbf53b9103cc0a117661f075efc08e4f67bbf29f9258d69be974511182284f9d787a067698fe687ffe17f0b341a461e5e13c056eeee9ad
Type fulltextMimetype application/pdf

Other links

Publisher's full text

Authority records

Ioannidis, IoannisNascetti, AndreaCeccato, Vania

Search in DiVA

By author/editor
Ioannidis, IoannisNascetti, AndreaCeccato, VaniaRobert, Haining
By organisation
Urban and Regional Studies
Geosciences, Multidisciplinary

Search outside of DiVA

GoogleGoogle Scholar
Total: 32 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

doi
urn-nbn

Altmetric score

doi
urn-nbn
Total: 261 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf