Applying Crime Prediction Techniques to Japan: A Comparison Between Risk Terrain Modeling and Other Methods

Ohyama, Tomoya; Amemiya, Mamoru

doi:10.1007/s10610-018-9378-1

Cited by 41 publications

(22 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Interpreting the plot in the top right corner, it can be interpreted similarly to an ROC curve, but the Y-axis is weighted according to the cumulative number of crimes (Mohler and Porter 2018;Ohyama and Amemiya 2018). This again illustrates that simply using the prior crime counts provides reasonable predictions for future data at the very highest crime locations, as well as provides additional evidence for the law of crime concentration (Weisburd 2015).…”

Section: [Insert Table 2 and Figure 1 Here]mentioning

confidence: 93%

Mapping the Risk Terrain for Crime Using Machine Learning

Wheeler

Steenbeek

2020

J Quant Criminol

View full text Add to dashboard Cite

Objectives: We illustrate how a machine learning algorithm, Random Forests, can provide accurate long-term predictions of crime at micro places relative to other popular techniques. We also show how recent advances in model summaries can help to open the 'black box' of Random Forests, considerably improving their interpretability. Methods: We generate long-term crime forecasts for robberies in Dallas at 200 by 200 feet grid cells that allow spatially varying associations of crime generators and demographic factors across the study area. We then show how using interpretable model summaries facilitate understanding the model's inner workings. Results: We find that Random Forests greatly outperform Risk Terrain Models and Kernel Density Estimation in terms of forecasting future crimes using different measures of predictive accuracy, but only slightly outperform using prior counts of crime. We find different factors that predict crime are highly non-linear and vary over space. Conclusions: We show how using black-box machine learning models can provide accurate micro placed based crime predictions, but still be interpreted in a manner that fosters understanding of why a place is predicted to be risky.

show abstract

Section: [Insert Table 2 and Figure 1 Here]mentioning

confidence: 93%

Mapping the Risk Terrain for Crime Using Machine Learning

Wheeler

Steenbeek

2020

J Quant Criminol

View full text Add to dashboard Cite

show abstract

“…To do this we use the predictions generated by RTM based on the 2013 through 2017 data, and then evaluate them against homeless related crimes (restricted to the same property and violent crime categories) in 2018. Figure 4 displays the proportion of area under the study on the X axis, and the Y axis displays the proportion of 2018 homeless crimes captured based on the combined RTM risk score (Ohyama & Amemiya, 2018). So the graph can be read as the top 5% of the RTM risk scores captures approximately 70% of the homeless related crime in LA.…”

Section: Referring Back Tomentioning

confidence: 99%

Using risk terrain modeling to predict homeless related crime in Los Angeles, California

Yoo

Wheeler

2019

Applied Geography

View full text Add to dashboard Cite

We apply Risk Terrain Modeling (RTM) to identify the factors that predict homeless related crime at micro grid cells in Los Angeles, CA. We find that place based factors predicting whether homeless individuals are victimized or the offender being homeless are largely consistent with one another. Out of 26 different crime attractors and generators, prior drug arrests, homeless shelters, and bus stops are the three biggest factors in predicting homeless related crime. We show how the RTM model can effectively forecast future homeless related crimes as well. This suggests that targeted spatial strategies can reduce both homeless offending and victimization risk. Given that the majority of homeless individuals are only intermittently homeless, place based strategies may be a more an effective way to limit risk than strategies that focus on individuals.

show abstract

“…The last two papers of this special issue introduce some reflections about the challenges faced by researchers and agencies in applying evidence-based crime forecasting approaches. Specifically, the study by Ohyama and Amemiya (2018) focuses on the Japanese situation where very low crime numbers seriously jeopardize the use of forecasting methods based on previous crime data. Comparing the performances of several crime forecasting techniques applied to thefts from vehicles in Fukuoka, Japan this paper proves that RTM outperforms other methods, thus supporting a predicting approach based on environmental factors.…”

Section: This Special Issuementioning

confidence: 99%

Forecasting Crime and Understanding its Causes. Applying Risk Terrain Modeling Worldwide

Kennedy

Dugato

2018

Eur J Crim Policy Res

View full text Add to dashboard Cite

We are pleased to introduce a special issue on risk terrain modeling (RTM) with expert contributions from around the world. The intent of this issue is to highlight the advances that have been made in the use of the RTM approach and to provide an opportunity to test its applicability in many different settings and geographies. The issue is also intended to provide insights for scholars who are interested in exploring the theoretical and methodological applications of RTM in relation to the unifying themes of environmental criminology and the spatial aspects of crime. Our intention was to address the different aspects of the inquiry inspired by the development of RTM, including theoretical tests, empirical verification, and practical applications. In addressing their research questions, authors were encouraged to present the rationale behind RTM, looking at contextual factors for making crime forecasts that are more useful for guiding interventions and more capable in identifying future possible crime concentrations. We added to this a planned solicitation of papers from researchers working in different countries, faced with diverse data issues, law enforcement practices, and analytical challenges. Indeed, the contributions of this special issue cover four continents,

show abstract

Applying Crime Prediction Techniques to Japan: A Comparison Between Risk Terrain Modeling and Other Methods

Cited by 41 publications

References 32 publications

Mapping the Risk Terrain for Crime Using Machine Learning

Mapping the Risk Terrain for Crime Using Machine Learning

Using risk terrain modeling to predict homeless related crime in Los Angeles, California

Forecasting Crime and Understanding its Causes. Applying Risk Terrain Modeling Worldwide

Contact Info

Product

Resources

About