BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Large Scale Ubiquitous Data Sources for Crime Prediction - Cristin
 a Kadar\, ETH Zurich
DTSTART:20180220T140000Z
DTEND:20180220T150000Z
UID:TALK97090@talks.cam.ac.uk
CONTACT:Alexander Vetterl
DESCRIPTION:In this talk\, I will present two approaches to geographical c
 rime profiling that leverage machine learning techniques and large scale u
 biquitous data sources. I will briefly touch on their motivation in crimin
 ology and urban studies\, as well as on their challenges and limitations.\
 n\nThe first work mines large-scale human mobility data to craft an extens
 ive set of features for yearly crime counts prediction in New York City. T
 raditional crime models based on census data are limited\, as they fail to
  capture the complexity and dynamics of human activity. With the rise of u
 biquitous computing\, there is the opportunity to improve such models with
  data that make for better proxies of human presence in cities. Our study 
 shows that spatial and spatio-temporal features derived from Foursquare ve
 nues and checkins\, subway rides\, and taxi rides\, improve the baseline m
 odels relying only on census data. The proposed ensemble machine learning 
 models achieve absolute R2 metrics of up to 65% (on a geographical out-of-
 sample test set) and up to 89% (on a temporal out-of-sample test set). Thi
 s proves that\, next to the residential population of an area\, the ambien
 t population there is strongly predictive of the area’s crime levels. We
  deep-dive into the main crime categories\, and find that the predictive g
 ain of the human dynamics features varies across crime types: such feature
 s bring the biggest boost in case of grand larcenies\, whereas assaults ar
 e already well predicted by the census features. Furthermore\, we identify
  and discuss top predictive features for the main crime categories. These 
 results offer valuable insights for those responsible for urban policy. \n
 \nThe second work investigates a forecasting approach for daily burglary r
 isk within a region of Switzerland characterized by significantly lower le
 vels of urbanization compared to the areas analyzed in prevailing crime pr
 ediction research. The lower levels of urbanization in combination with hi
 gh spatial and temporal granularity pose a significant challenge to buildi
 ng accurate prediction models necessary to derive feasible and effective p
 reventive actions\, e.g. in form of police patrols. We employ machine lear
 ning methods\, which allow for integration of diverse fine-grained data on
  the demographic\, geographic\, economic\, temporal\, and meteorological c
 haracteristics of the environment\, next to past burglary events. We propo
 se an approach which addresses the sparsity of the data and significantly 
 outperforms the baseline implementation of a prospective hotspot model\, w
 hich only makes use of historical crime data and is an industry standard. 
 For instance\, by setting the coverage of the predicted areas to 5% of the
  total studied area\, the model is able to predict the committed burglarie
 s on a specific day within a four-hectare rectangular area with an average
  hit ratio of 57% compared to the 36% hit ratio of the baseline. This rese
 arch has direct implications for decision makers in charge of resource all
 ocation for crime prevention.\n\nBio:\nCristina is a PhD candidate at the 
 Department of Management\, Technology\, and Economics (D-MTEC) of the Swis
 s Federal Institute of Technology in Zurich (ETH Zurich). She holds a M.Sc
 . with Honors in Software Engineering from Technical University of Munich 
 and a B.Sc. in Computer Science from Leibniz University of Hanover. Her re
 search interests evolve around information systems\, computational social 
 science\, and applied machine learning\, with a focus on crime and fear of
  crime.\n
LOCATION:LT2\, Computer Laboratory\, William Gates Building
END:VEVENT
END:VCALENDAR