Actionable Insights on Philadelphia Crime Hot-Spots: Clustering and Statistical Analysis to Inform Future Crime Legislation
Authors:
Ishan S. Khare,
Tarun K. Martheswaran,
Rahul K. Thomas,
Aditya Bora
Abstract:
Philadelphia's problem with high crime rates continues to be exacerbated as Philadelphia's residents, community leaders, and law enforcement officials struggle to address the root causes of the problem and make the city safer for all. In this work, we deeply understand crime in Philadelphia and offer novel insights for crime mitigation within the city. Open source crime data from 2012-2022 was obt…
▽ More
Philadelphia's problem with high crime rates continues to be exacerbated as Philadelphia's residents, community leaders, and law enforcement officials struggle to address the root causes of the problem and make the city safer for all. In this work, we deeply understand crime in Philadelphia and offer novel insights for crime mitigation within the city. Open source crime data from 2012-2022 was obtained from OpenDataPhilly. Density-Based Spatial Clustering of Applications with Noise (DBSCAN) was used to cluster geographic locations of crimes. Clustering of crimes within each of 21 police districts was performed, and temporal changes in cluster distributions were analyzed to develop a Non-Systemic Index (NSI). Home Owners' Loan Corporation (HOLC) grades were tested for associations with clusters in police districts labeled `systemic.' Crimes within each district were highly clusterable, according to Hopkins' Mean Statistics. NSI proved to be a good measure of differentiating systemic ($<$ 0.06) and non-systemic ($\geq$ 0.06) districts. Two systemic districts, 19 and 25, were found to be significantly correlated with HOLC grade (p $=2.02 \times 10^{-19}$, p $=1.52 \times 10^{-13}$). Philadelphia crime data shows a high level of heterogeneity between districts. Classification of districts with NSI allows for targeted crime mitigation strategies. Policymakers can interpret this work as a guide to interventions.
△ Less
Submitted 28 June, 2023;
originally announced June 2023.
Compressed Sensing using Generative Models
Authors:
Ashish Bora,
Ajil Jalal,
Eric Price,
Alexandros G. Dimakis
Abstract:
The goal of compressed sensing is to estimate a vector from an underdetermined system of noisy linear measurements, by making use of prior knowledge on the structure of vectors in the relevant domain. For almost all results in this literature, the structure is represented by sparsity in a well-chosen basis. We show how to achieve guarantees similar to standard compressed sensing but without employ…
▽ More
The goal of compressed sensing is to estimate a vector from an underdetermined system of noisy linear measurements, by making use of prior knowledge on the structure of vectors in the relevant domain. For almost all results in this literature, the structure is represented by sparsity in a well-chosen basis. We show how to achieve guarantees similar to standard compressed sensing but without employing sparsity at all. Instead, we suppose that vectors lie near the range of a generative model $G: \mathbb{R}^k \to \mathbb{R}^n$. Our main theorem is that, if $G$ is $L$-Lipschitz, then roughly $O(k \log L)$ random Gaussian measurements suffice for an $\ell_2/\ell_2$ recovery guarantee. We demonstrate our results using generative models from published variational autoencoder and generative adversarial networks. Our method can use $5$-$10$x fewer measurements than Lasso for the same accuracy.
△ Less
Submitted 9 March, 2017;
originally announced March 2017.
