Transaction Search Form: please type in any of the fields below.
Date: August 17, 2024 Sat
Time: 2:50 am
Time: 2:50 am
Results for crime prediction
4 results foundAuthor: Hunt, Priscillia Title: Evaluation of the Shreveport Predictive Policing Experiment Summary: Predictive policing is the application of statistical methods to identify likely targets for police intervention (the predictions) to prevent crimes or solve past crimes, followed by conducting interventions against those targets. The concept has been of high interest in recent years as evidenced by the growth of academic, policy, and editorial reports; however, there have been few formal evaluations of predictive policing efforts to date. In response, the National Institute of Justice (NIJ) funded the Shreveport Police Department (SPD) in Louisiana to conduct a predictive policing experiment in 2012. SPD staff developed and estimated a statistical model of the likelihood of property crimes occurring within block-sized areas. Then, using a blocked randomized approach to identify treatment and control district pairs, districts assigned to the treatment group were given maps that highlighted blocks predicted to be at higher risk of property crime. These districts were also provided with overtime resources to conduct special operations. Control districts conducted property crime-related special operations using overtime resources as well, just targeting areas that had recently seen property crimes (hot spots). This study presents results of an evaluation of the processes in addition to the impacts and costs of the SPD predictive policing experiment. It should be of interest to those considering predictive policing and directed law enforcement systems and operations, and to analysts conducting experiments and evaluations of public safety strategies. This evaluation is part of a larger project funded by the NIJ, composed of two phases. Phase I focuses on the development and estimation of predictive models, and Phase II involves implementation of a prevention model using the predictive model. For Phase II, RAND is evaluating predictive policing strategies conducted by the SPD and the Chicago Police Department (contract #2009-IJ-CX-K114). This report is one product from Phase II. Details: Santa Monica, CA: RAND, 2014. 88p. Source: Internet Resource: Accessed August 4, 2014 at: http://www.rand.org/content/dam/rand/pubs/research_reports/RR500/RR531/RAND_RR531.pdf Year: 2014 Country: United States URL: http://www.rand.org/content/dam/rand/pubs/research_reports/RR500/RR531/RAND_RR531.pdf Shelf Number: 132885 Keywords: Crime AnalysisCrime PredictionCrime PreventionHot-Spots PolicingPredictive PolicingProperty Crimes |
Author: Ratcliffe, Jerry H. Title: Predictive Modeling Combining Short and Long-Term Crime Risk Potential: Final Report Summary: This research team (Temple University and industry partner Azavea) developed a technology capable of predicting future crime risk potential based on a number of grounded theoretical approaches to understanding localized spatial crime patterns. With regard to long-term crime risk changes, a stable crime niche model assumes that communities occupy crime niches in a broader jurisdiction, niches that are largely stable from year to year and have self-maintaining properties. Thus crime in one year may be predicted best by crime from the previous year. Alternatively, a structural model assumes that key current demographic conditions, such as socioeconomic status and racial composition, generally shape crime levels. Finally, a dynamic ecological and structural model assumes, net of the connections between current crime and demographic structure, that current structural conditions influence future long term changes in crime for a year in the future. The focus here is on ecological crime discontinuities, with priority assigned to demographic factors shaping such crime shifts over time. At the same time, ecological crime continuities also are present to a degree, linking current and future crime levels. These models were compared in the research study. The research team also examined what role near-repeat crime events, indicative of a short-term change in relative risk, have in modifying this relationship. Near repeats occur when a crime influences the likelihood of another crime within a narrow space and time window after the originator event. In particular, the 'boost' hypothesis (also known as 'event dependency') suggests that subsequent events are conditional on the originator event because (for example) the same offender returns to the area, or there is a retaliatory event. Using 2009 and 2010 reported crime for the City of Philadelphia, PA (USA) we identified that the demographics-plus-crime was the most parsimonious and accurate for robbery, burglary, aggravated assault, and vehicle theft when predicted from year-to-year in small geographic areas of 500 feet by 500 feet grid cells. Lower volume crime types (homicide and rape) were predicted as well as, or better, by the demographics-only model. We then added an event-dependency risk surface to the long-term crime risk predictions and estimated what impact this near repeat surface played in changing the accuracy and parsimony of the crime prediction. The best combination of accuracy and model parsimony was estimated by comparing differences in Bayesian Information Criterion (BIC) values. Near repeat patterns were estimated for two week periods across spatial bands of 250 feet width. These near repeat patterns were translated to a mapped risk surface and added to the long-term risk prediction surfaces. In this part of the study, 2012 crime was used to predict 2013 crime in the City of Philadelphia for two of the most frequent types of part 1 crime: robbery and burglary. With repeated examination of two-week predictions across 500 foot square grid cells, the strongest BIC value was identified with a model that combines crime from the previous year, change in demographic structure, and an adjustment for the near repeat phenomenon. Mixed effects logit models suggest that long-term (year-on-year) crime and demographic changes are more influential in this model than near repeats. Theoretically, this means that long term ecological crime continuities, long term crime discontinuities arising from stratification patterns in class and race, and near-term crime continuities in time and space all shape the two week, micro-scale predictions. In summary, a model combining community structural characteristics, crime counts from the previous year, and an estimate of near repeat activity generated the best results overall. This tells us that small scale, short term crime occurrences reflect a complex mix of near-term crime continuities, ecological crime continuities, and ecological structure which generates ecological crime discontinuities forward in time. The industry partner, Azavea, has created a free software program (PROVE) to perform these calculations for state and municipal police departments. Details: Philadelphia: Temple University, Center for Security and Crime Science, 2016. 131p. Source: Internet Resource: Accessed July 26, 2016 at: https://www.ncjrs.gov/pdffiles1/nij/grants/249934.pdf Year: 2016 Country: United States URL: https://www.ncjrs.gov/pdffiles1/nij/grants/249934.pdf Shelf Number: 139857 Keywords: Computer TechnologyCrime AnalysisCrime Prediction |
Author: Willits, Dale Title: Situational and Individual Predictors of Violent Intentions: A Factorial Survey Approach Summary: Though many criminological perspectives suggest that violence is the result of both individual and situational factors, the majority of criminological research focuses narrowly on individual-level factors. The current study contributes to the literature by utilizing a factorial survey design to examine both the independent and interactive effects of situational and individual predictors of violent behavioral intentions. This factorial survey presented college respondents with randomly generated versions of a hypothetical situation depicting interpersonal conflict and also gathered data about a variety of individual level factors known to predict violence. In order to improve the validity of the factorial survey method, the vignettes utilized in this study were based on those utilized in prior research and were pretested in a series of focus groups. The factorial elements of the vignette were inspired by psychological and qualitative sociological research on violence and aggression. Utilizing a sample size of 751 respondents, I estimate a series of multilevel regression models predicting violent behavioral intentions. Results suggest that both individual level and situational factors are important predictors of violent intentions. Specifically, physical provocation, the attention of an audience, and the presence of aggressive cues all significantly predicted violent intentions. Results also suggest that, in addition to their separate relationships with violent intentions, individual and situational factors interact to predict violent intentions. After demonstrating the importance of situational factors in predicting violent intentions, I then demonstrate the utility of a situational perspective to criminology more broadly by providing situational tests of general strain theory and situational action theory. These situational tests demonstrate general support for both theoretical perspectives and highlight the importance of utilizing the situation as the unit of analysis for studying micro-social processes. Details: Albuquerque: University of New Mexico, 2012. 163p. Source: Internet Resource: Dissertation: Accessed February 1, 2017 at: http://repository.unm.edu/bitstream/handle/1928/21091/Willits%20Dissertation%20Final%20Version.pdf?sequence=1&isAllowed=y Year: 2012 Country: United States URL: http://repository.unm.edu/bitstream/handle/1928/21091/Willits%20Dissertation%20Final%20Version.pdf?sequence=1&isAllowed=y Shelf Number: 140785 Keywords: Crime PredictionViolence Violence Prediction |
Author: Dugato, Marco Title: Prevedere i Furti in Abitazione Summary: Introduction - This research is the product of a path that the Transcrime center has started in 2007 with the goal of develop models for the analysis of the risk and for the prevention of crime. - In 2008, Transcrime presented the first study, promoted by the then Prefect of Naples Alessandro Pansa, on the spatial analysis of crime in San Lorenzo district. This study he showed how the crimes were concentrated in time and space. - In the meantime, technological development has supported the progress of knowledge criminological in the analysis and in the prediction of criminal behavior. - Today this research presents a model predictive for thefts in the home that is the fruit of a collaboration between Transcrime and the Ministry of the Interior, Department of Public Security. - In the following pages we indicate because the study of home burglaries is relevant at European and Italian level. Yes then explains how to use models forecasting can help prevent home burglaries. - Finally, head and yes positively applies a model forecast to the cities of Milan, Rome and Bari, suggesting hypotheses for the reduction of the phenomenon. Details: Milano, Italy: Transcrime, 2015. 15p. Source: Internet Resource (in Italian): Accessed January 16, 2019 at: http://www.transcrime.it/en/pubblicazioni/transcrime-research-in-brief-serie-italia/ Year: 2015 Country: Italy URL: http://www.transcrime.it/wp-content/uploads/2015/04/Research-in-Brief.pdf Shelf Number: 154220 Keywords: Crime PredictionCrime PreventionHome BurglariesItalyPrediction ModelsTechnological DevelopmentTechnologiesTheft |