Abstract
Predicting Burglaries and Other Incidents (PBOI) is a project to estimate the possibilities of burglary at a given location at a specified time duration. Crime prediction using fixed-location observation points are being used by police. Prediction using location-independent methods, which use a variety of environmental and other additional sources (data), is currently being explored by researchers.
PBOI is a pilot-project to study the feasibility of estimating burglaries using machine learning methods, which uses data from Interpolis/Achmea, Dutch police, TNO and a number of open data sources.
PBOI uses open-source tools to import, analyze, generate a model (using data), and generate predictions. The model is generated using a machine learning algorithm. The algorithm is used to model systems from historic data, which can be mapped to non-parametric functions with independent variables.
Based on the findings and a comparative study, the special algorithm performs better than others on the crime prediction domain. The predictions are populated on a map, which can help police and insurance professionals, to make informed decisions and to avoid burglaries and inform clients respectively.
PBOI is a pilot-project to study the feasibility of estimating burglaries using machine learning methods, which uses data from Interpolis/Achmea, Dutch police, TNO and a number of open data sources.
PBOI uses open-source tools to import, analyze, generate a model (using data), and generate predictions. The model is generated using a machine learning algorithm. The algorithm is used to model systems from historic data, which can be mapped to non-parametric functions with independent variables.
Based on the findings and a comparative study, the special algorithm performs better than others on the crime prediction domain. The predictions are populated on a map, which can help police and insurance professionals, to make informed decisions and to avoid burglaries and inform clients respectively.
Original language | English |
---|---|
Supervisors/Advisors |
|
Award date | 25 Sept 2015 |
Place of Publication | Eindhoven |
Publisher | |
Print ISBNs | 978-90-444-1379-3 |
Publication status | Published - 25 Sept 2015 |