Work

Geospatial case studies

Mass Shooting Spatial Intelligence Dashboard (2024)

This interactive geospatial dashboard visualizes mass shooting incidents across the United States using data from Gun Violence Archive. The project applies spatial analysis techniques to identify geographic concentrations, regional clustering patterns, and high-risk zones.

Tools Used:

  • QGIS
  • qgis2web (github)
  • Spatial clustering analysis
  • Policy intelligence integration

Purpose:
To provide a Common Operating Picture (COP) for policymakers to allocate prevention resources strategically.

Users can:

  • Click incidents to view casualties and dates
  • Zoom to state-level concentration

  • Identify high-density regions

  • Support evidence-based policy recommendations

Spatiotemporal Crime Hot Spot Analysis After a Major Disaster

Project Overview

Following a simulated major natural disaster, I was tasked with analyzing robbery incidents across the City of Philadelphia to determine how crime patterns evolved over time. The objective was to identify statistically significant hot spots and detect areas where crime intensified or diminished, enabling spatially targeted crime prevention strategies.

This analysis integrates both spatial and temporal dimensions using advanced GIS statistical tools.

Problem Statement

Crime incidents are not randomly distributed in space or time. Following disasters, opportunistic criminal activity may increase in vulnerable neighborhoods. Decision-makers require precise spatial intelligence to allocate law enforcement resources effectively.

This project answers:

  • Where are robbery incidents clustering?
  • Which areas are intensifying over time?
  • Which hot spots are persistent versus emerging?
  • Where should targeted intervention occur?

Emerging Hot Spot Classification Output

Emerging Hot Spot Analysis output showing statistically significant spatiotemporal robbery clusters categorized by pattern type.

Results

The analysis revealed:

  • Concentrated intensifying robbery hot spots in specific urban corridors
  • Persistent hot spots indicating long-term crime concentration
  • Emerging clusters suggesting post-disaster vulnerability
  • Cold spots where robbery declined significantly
  • These findings provide actionable intelligence for law enforcement resource allocation and targeted intervention planning.

Exploring Crime Trends in Chicago Using R

Project Overview

This project investigates spatial patterns of homicide and burglary incidents in Chicago using publicly available data from the Chicago Police Department’s CLEAR (Citizen Law Enforcement Analysis and Reporting) system.

The objective was to determine whether crime incidents were clustered, dispersed, or randomly distributed across the city using advanced spatial statistical techniques in R.

 

Research Objective

Crime data often exhibits spatial dependence. Understanding these spatial patterns allows policymakers and law enforcement agencies to:

  • Identify crime concentration zones
  • Evaluate clustering intensity
  • Detect statistically significant deviations from randomness
  • Improve resource allocation strategies

This project applies hypothesis-driven spatial statistical testing to examine the distribution of crime in Chicago.

Data Collection & Preparation

  • Source: Chicago Police Department CLEAR system
  • Extracted crime types:
    • Homicide
    • Burglary
    • Motor Vehicle Theft

 

Data preprocessing included:

  • Removal of unnecessary attributes
  • Cleaning inconsistent records
  • Exporting structured .csv datasets
  • Formatting spatial coordinates for analysis in R

 

Methodology

The analysis applied the following spatial statistical methods:

Spatial Summary Statistics

  • Mean Center
  • Standard Deviation
  • Nearest Neighbor Analysis

  • Kernel Density Estimation (KDE)

 

Hypothesis Testing (CSR vs. Spatial Pattern)

  • Ripley’s K Function (Lhat) with Monte Carlo simulation
  • Spatial Autocorrelation (Moran’s I)

The central hypothesis tested whether crime incidents followed Complete Spatial Randomness (CSR) or exhibited statistically significant clustering or dispersion.

Key Findings

  • Kernel Density Estimation identified high-density crime corridors.

  • Ripley’s K Function analysis indicated significant deviations from CSR.

  • Homicide incidents showed measurable spatial structure rather than randomness.
  • Some burglary distributions showed dispersion relative to expected patterns.

The statistical tests confirmed that Chicago crime patterns are spatially structured and scale-dependent.

 

Strategic Implications

  • Spatial clustering supports hotspot policing strategies
  • Identifying dispersion zones aids in anomaly detection
  • Statistical validation strengthens evidence-based crime policy
  • Multi-scale analysis improves understanding of neighborhood-level risk

This project demonstrates the integration of statistical programming and geospatial intelligence.

 

Tools & Technologies

  • RStudio
    • spatstat package
    • Monte Carlo simulation
    • Kernel Density Estimation
    • Ripley’s K Function

    • Spatial Autocorrelation

A Geospatial Analysis of the 2020 Beirut Explosion

Mapping Destruction and Recovery: Beirut Port Explosion

BOTTOM LINE UP FRONT (BLUF)

In the aftermath, national and international attention turned to the root causes of poor governance and the lack of enforcement around hazardous material management. The analysis demonstrates Lebanon's lack of preparedness across all aspects of disaster management.


On August 4, 2020, at approximately 6:08 PM local time, a massive explosion rocked the Port of Beirut, Lebanon's primary seaport and a densely populated area near the city center. The initial fire broke out in Warehouse 12 at the Port of Beirut, causing smoke and small explosions. Shortly after, a massive explosion occurred, producing a blast wave that shattered windows, destroyed buildings, and caused destruction across a radius of several kilometers.


The source of the explosion was approximately 2,750 metric tons of improperly stored ammonium nitrate in a port warehouse since 2014. The blast killed over 220 people, injuring more than 7000 and displacing approximately 300,000 residents. The powerful shockwave generated by the explosion significantly impacted urban infrastructure, transportation networks, and environmental
conditions throughout the city.

AIRBORNE TOXIC POLLUTANTS

The explosion released significant air pollutants, notably nitrogen dioxide (NO₂), a key indicator of air quality. While NO₂ occurs naturally, elevated levels from anthropogenic sources can pose serious health risks with prolonged exposure. Using Sentinel-5P satellite data, this study mapped atmospheric NO₂ concentrations before and after the blast. Findings revealed a sharp spike immediately following the explosion, with levels returning to normal within six days. The following analysis highlights the short-term impact of the disaster on Beirut and its surroundings' air quality.

Follow us