Geosocial media data as predictors in a GWR application to forecast crime hotspots

Alina Ristea; O. Kounadi; Michael Leitner

doi:10.4230/LIPIcs.GIScience.2018.56

Geosocial media data as predictors in a GWR application to forecast crime hotspots

Alina Ristea
, O. Kounadi
, Michael Leitner

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

7 Citations (Scopus)

572 Downloads (Pure)

Abstract

In this paper we forecast hotspots of street crime in Portland, Oregon. Our approach uses geosocial media posts, which define the predictors in geographically weighted regression (GWR) models. We use two predictors that are both derived from Twitter data. The first one is the population at risk of being victim of street crime. The second one is the crime related tweets. These two predictors were used in GWR to create models that depict future street crime hotspots. The predicted hotspots enclosed more than 23% of the future street crimes in 1% of the study area and also outperformed the prediction efficiency of a baseline approach. Future work will focus on optimizing the prediction parameters and testing the applicability of this approach to other mobile crime types.

Original language	English
Title of host publication	10th International Conference on Geographic Information Science, GIScience 2018
Editors	Sester M Griffin A.L. Winter S.
Publisher	Dagstuhl
ISBN (Print)	9783959770835
DOIs	https://doi.org/10.4230/LIPIcs.GIScience.2018.56
Publication status	Published - 1 Aug 2018
Event	10th International Conference on Geographic Information Science, GIScience 2018 - RMIT University, Melbourne, Australia Duration: 28 Aug 2018 → 31 Aug 2018 Conference number: 10

Publication series

Name	Leibniz International Proceedings in Informatics, LIPIcs
Volume	114

Conference

Conference	10th International Conference on Geographic Information Science, GIScience 2018
Abbreviated title	GIScience 2018
Country/Territory	Australia
City	Melbourne
Period	28/08/18 → 31/08/18

Keywords

Geographically weighted regression
Geosocial media
Population at risk
Spatial crime prediction
Street crime
ITC-GOLD

Access to Document

10.4230/LIPIcs.GIScience.2018.56Licence: CC BY

ArticleFinal published version, 383 KBLicence: CC BY

Cite this

Ristea, A., Kounadi, O., & Leitner, M. (2018). Geosocial media data as predictors in a GWR application to forecast crime hotspots. In S. M. Griffin A.L. Winter S. (Ed.), 10th International Conference on Geographic Information Science, GIScience 2018 (Leibniz International Proceedings in Informatics, LIPIcs; Vol. 114). Dagstuhl. https://doi.org/10.4230/LIPIcs.GIScience.2018.56

@inproceedings{dce8c68e68df4b35a7901e11bd0fce9f,

title = "Geosocial media data as predictors in a GWR application to forecast crime hotspots",

abstract = "In this paper we forecast hotspots of street crime in Portland, Oregon. Our approach uses geosocial media posts, which define the predictors in geographically weighted regression (GWR) models. We use two predictors that are both derived from Twitter data. The first one is the population at risk of being victim of street crime. The second one is the crime related tweets. These two predictors were used in GWR to create models that depict future street crime hotspots. The predicted hotspots enclosed more than 23\% of the future street crimes in 1\% of the study area and also outperformed the prediction efficiency of a baseline approach. Future work will focus on optimizing the prediction parameters and testing the applicability of this approach to other mobile crime types.",

keywords = "Geographically weighted regression, Geosocial media, Population at risk, Spatial crime prediction, Street crime, ITC-GOLD",

author = "Alina Ristea and O. Kounadi and Michael Leitner",

year = "2018",

month = aug,

day = "1",

doi = "10.4230/LIPIcs.GIScience.2018.56",

language = "English",

isbn = "9783959770835",

series = "Leibniz International Proceedings in Informatics, LIPIcs",

publisher = "Dagstuhl",

editor = "\{Griffin A.L. Winter S.\}, \{Sester M\}",

booktitle = "10th International Conference on Geographic Information Science, GIScience 2018",

address = "Germany",

note = "10th International Conference on Geographic Information Science, GIScience 2018, GIScience 2018 ; Conference date: 28-08-2018 Through 31-08-2018",

}

Ristea, A, Kounadi, O & Leitner, M 2018, Geosocial media data as predictors in a GWR application to forecast crime hotspots. in SM Griffin A.L. Winter S. (ed.), 10th International Conference on Geographic Information Science, GIScience 2018. Leibniz International Proceedings in Informatics, LIPIcs, vol. 114, Dagstuhl, 10th International Conference on Geographic Information Science, GIScience 2018, Melbourne, Victoria, Australia, 28/08/18. https://doi.org/10.4230/LIPIcs.GIScience.2018.56

Geosocial media data as predictors in a GWR application to forecast crime hotspots. / Ristea, Alina; Kounadi, O.; Leitner, Michael.
10th International Conference on Geographic Information Science, GIScience 2018. ed. / Sester M Griffin A.L. Winter S. Dagstuhl, 2018. (Leibniz International Proceedings in Informatics, LIPIcs; Vol. 114).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Geosocial media data as predictors in a GWR application to forecast crime hotspots

AU - Ristea, Alina

AU - Kounadi, O.

AU - Leitner, Michael

N1 - Conference code: 10

PY - 2018/8/1

Y1 - 2018/8/1

N2 - In this paper we forecast hotspots of street crime in Portland, Oregon. Our approach uses geosocial media posts, which define the predictors in geographically weighted regression (GWR) models. We use two predictors that are both derived from Twitter data. The first one is the population at risk of being victim of street crime. The second one is the crime related tweets. These two predictors were used in GWR to create models that depict future street crime hotspots. The predicted hotspots enclosed more than 23% of the future street crimes in 1% of the study area and also outperformed the prediction efficiency of a baseline approach. Future work will focus on optimizing the prediction parameters and testing the applicability of this approach to other mobile crime types.

AB - In this paper we forecast hotspots of street crime in Portland, Oregon. Our approach uses geosocial media posts, which define the predictors in geographically weighted regression (GWR) models. We use two predictors that are both derived from Twitter data. The first one is the population at risk of being victim of street crime. The second one is the crime related tweets. These two predictors were used in GWR to create models that depict future street crime hotspots. The predicted hotspots enclosed more than 23% of the future street crimes in 1% of the study area and also outperformed the prediction efficiency of a baseline approach. Future work will focus on optimizing the prediction parameters and testing the applicability of this approach to other mobile crime types.

KW - Geographically weighted regression

KW - Geosocial media

KW - Population at risk

KW - Spatial crime prediction

KW - Street crime

KW - ITC-GOLD

UR - https://www.scopus.com/pages/publications/85051342572

U2 - 10.4230/LIPIcs.GIScience.2018.56

DO - 10.4230/LIPIcs.GIScience.2018.56

M3 - Conference contribution

AN - SCOPUS:85051342572

SN - 9783959770835

T3 - Leibniz International Proceedings in Informatics, LIPIcs

BT - 10th International Conference on Geographic Information Science, GIScience 2018

A2 - Griffin A.L. Winter S., Sester M

PB - Dagstuhl

T2 - 10th International Conference on Geographic Information Science, GIScience 2018

Y2 - 28 August 2018 through 31 August 2018

ER -

Geosocial media data as predictors in a GWR application to forecast crime hotspots

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this