Solving burglary offences: building a model to predict clearance of burglary following initial investigation

Tom Olphin; Siddhartha Bandyopadhyay; Anindya Banerjee

doi:10.1093/police/paz002

Solving burglary offences: building a model to predict clearance of burglary following initial investigation

Tom Olphin, Siddhartha Bandyopadhyay, Anindya Banerjee

Economics

Research output: Contribution to journal › Article › peer-review

222 Downloads (Pure)

Abstract

This research identifies solvability factors for burglary offences, and develops previous research by building, and testing, an algorithmic prediction model for solvability of burglary offences. It is based on a full population of 9,655 burglary offences reported to a UK Police force between 1st April 2012 and 30th April 2015. The dataset was split in half, with half being used to build the model and half to test it. Thirty-one solvability factors were identified, along with nine case-limiting factors, allowing a logit model to be built to predict solvability of burglary offences. When compared to a model of investigating all burglary offences, use of this model would reduce investigative workload by up to 42.2%, or 1,321 cases per year.

Original language	English
Number of pages	16
Journal	Policing: A journal of Policy and Practice
Early online date	21 Feb 2019
DOIs	https://doi.org/10.1093/police/paz002
Publication status	E-pub ahead of print - 21 Feb 2019

Keywords

clearance
solvability
burglary
prediction
algorithmic prediction
predictive accuracy

Access to Document

10.1093/police/paz002Licence: None: All rights reserved

Banerjee_et_al_Solving_burglary_offences_Policing_2019
This is a pre-copyedited, author-produced PDF of an article accepted for publication in Policing: a Journal of Policy and Policing following peer review. The version of recordTom W L Olphin, Siddhartha Bandyopadhyay, Anindya Banerjee, Solving Burglary Offences: Building a Model to Predict Clearance of Burglary Following Initial Investigation, Policing: A Journal of Policy and Practice, is available online at: https://doi.org/10.1093/police/paz002.
Accepted author manuscript, 535 KBLicence: None: All rights reserved

https://academic.oup.com/policing/advance-article/doi/10.1093/police/paz002/5359499Licence: None: All rights reserved

Cite this

@article{0bd6d44dcca74da2b574617ac0350513,

title = "Solving burglary offences: building a model to predict clearance of burglary following initial investigation",

abstract = "This research identifies solvability factors for burglary offences, and develops previous research by building, and testing, an algorithmic prediction model for solvability of burglary offences. It is based on a full population of 9,655 burglary offences reported to a UK Police force between 1st April 2012 and 30th April 2015. The dataset was split in half, with half being used to build the model and half to test it. Thirty-one solvability factors were identified, along with nine case-limiting factors, allowing a logit model to be built to predict solvability of burglary offences. When compared to a model of investigating all burglary offences, use of this model would reduce investigative workload by up to 42.2%, or 1,321 cases per year.",

keywords = "clearance, solvability, burglary, prediction, algorithmic prediction, predictive accuracy",

author = "Tom Olphin and Siddhartha Bandyopadhyay and Anindya Banerjee",

year = "2019",

month = feb,

day = "21",

doi = "10.1093/police/paz002",

language = "English",

journal = "Policing: A journal of Policy and Practice",

issn = "1752-4512",

publisher = "Oxford University Press",

}

TY - JOUR

T1 - Solving burglary offences

T2 - building a model to predict clearance of burglary following initial investigation

AU - Olphin, Tom

AU - Bandyopadhyay, Siddhartha

AU - Banerjee, Anindya

PY - 2019/2/21

Y1 - 2019/2/21

N2 - This research identifies solvability factors for burglary offences, and develops previous research by building, and testing, an algorithmic prediction model for solvability of burglary offences. It is based on a full population of 9,655 burglary offences reported to a UK Police force between 1st April 2012 and 30th April 2015. The dataset was split in half, with half being used to build the model and half to test it. Thirty-one solvability factors were identified, along with nine case-limiting factors, allowing a logit model to be built to predict solvability of burglary offences. When compared to a model of investigating all burglary offences, use of this model would reduce investigative workload by up to 42.2%, or 1,321 cases per year.

AB - This research identifies solvability factors for burglary offences, and develops previous research by building, and testing, an algorithmic prediction model for solvability of burglary offences. It is based on a full population of 9,655 burglary offences reported to a UK Police force between 1st April 2012 and 30th April 2015. The dataset was split in half, with half being used to build the model and half to test it. Thirty-one solvability factors were identified, along with nine case-limiting factors, allowing a logit model to be built to predict solvability of burglary offences. When compared to a model of investigating all burglary offences, use of this model would reduce investigative workload by up to 42.2%, or 1,321 cases per year.

KW - clearance

KW - solvability

KW - burglary

KW - prediction

KW - algorithmic prediction

KW - predictive accuracy

U2 - 10.1093/police/paz002

DO - 10.1093/police/paz002

M3 - Article

SN - 1752-4512

JO - Policing: A journal of Policy and Practice

JF - Policing: A journal of Policy and Practice

ER -

Solving burglary offences: building a model to predict clearance of burglary following initial investigation

Abstract

Keywords

Access to Document

Fingerprint

Cite this