Enhancing short-term crime prediction with human mobility flows and deep learning architectures

Abstract

Place-based short-term crime prediction models leverage the spatio-temporal patterns of historical crimes to predict aggregate volumes of crime incidents at specific locations over time. Under the umbrella of the crime opportunity theory, that suggests that human mobility can play a role in crime generation, increasing attention has been paid to the predictive power of human mobility in place-based short-term crime models. Researchers have used call detail records (CDR), data from location-based services such as Foursquare or from social media to characterize human mobility; and have shown that mobility metrics, together with historical crime data, can improve short-term crime prediction accuracy. In this paper, we propose to use a publicly available fine-grained human mobility dataset from a location intelligence company to explore the effects of human mobility features on short-term crime prediction. For that purpose, we conduct a comprehensive evaluation across multiple cities with diverse demographic characteristics, different types of crimes and various deep learning models; and we show that adding human mobility flow features to historical crimes can improve the F1 scores for a variety of neural crime prediction models across cities and types of crimes, with improvements ranging from 2% to 7%. Our analysis also shows that some neural architectures can slightly improve the crime prediction performance when compared to non-neural regression models by at most 2%.

Crime Prediction Using Machine Learning: A Comparative Analysis

“…Some important themes have been identified upon the analysis of research studies and their comparison. forecasting for every grid cell at an hourly temporal The method is implemented using Kera on top of the Theano scale and the predictions themselves are accurate The authors verified that random forests delivered the Ippolito and Lozano (2020) enabled authors to compare the performance of various highest scores in the metrics of specificity, accuracy and ML algorithms precision for the fiscal data of 2018 [36] The authors used neural networks, random trees, and Bayesian Through the comparison of results offered by Wu et al (2022) networks and data mining for the analysis and prediction algorithms, the authors determine that random trees delivered of crime rules from the acquired data. Data from 2015 to 2018 better outcomes and results than that of Bayesian Networks was used for the YD county and neural networks [37] An SBCPM or assemble-stacking-based crime prediction…”

“…In addition, these weights are primarily adjusted to random values. Thus, in the sequence, these weights are adjusted by an iterative process, which helps minimize errors (Wu et al, 2022).…”

A Systematic Review of Mobile Phone Data in Crime Applications: A Coherent Taxonomy Based on Data Types and Analysis Perspectives, Challenges, and Future Research Directions

“…For example, Bogomolov et al [13] extracted human mobility patterns from mobile phone records to predict crime hotspots in London by using the Random Forest classifier to classify geographical areas into two classes based on whether they displayed high or low crime levels. However, Wu et al [96] criticized previous data collection methods such as CDRs, Twitter, and Foursquare data in terms of errors in estimating mobility flows for crime prediction, choosing instead to estimate human origin-destination mobility flows using GPS data alongside applied deep learning models such as the gated recurrent units (GRU) model and the graph convolution network (GCN).…”