Module 1 Blog Post: Crime Analysis and Hotspot Mapping in ArcGIS Pro

This week's lab provided an excellent opportunity to compare several commonly used crime mapping techniques while reinforcing many of the spatial analysis tools available in ArcGIS Pro. Although each exercise built on concepts introduced earlier in the course, this lab demonstrated how different analytical approaches can influence the way crime patterns are interpreted and how those results might be used to support decision-making.

The first part of the exercise focused on creating a choropleth map displaying burglary rates throughout Washington, DC. Rather than simply symbolizing raw crime counts, this map represented burglary rates by census tract, making it easier to compare areas of different population sizes. I spent time adjusting the classification method, color ramp, and map layout to create a product that was both accurate and visually balanced. Adding the required cartographic elements—including the title, legend, north arrow, scale bar, projection, author information, and data sources—reinforced the importance of creating maps that effectively communicate information to an audience instead of simply displaying data.

---

The second portion of the lab introduced Kernel Density analysis using assault incidents. This technique creates a continuous surface that estimates where crime events are most concentrated by assigning greater influence to incidents located near the center of the search radius. After generating the density raster, I experimented with several classification methods and color ramps before selecting a six-class Natural Breaks (Jenks) classification that clearly emphasized the highest-density areas. This process demonstrated how cartographic choices can significantly improve the readability and interpretation of a map.

---

The final portion of the lab compared three different hotspot analysis techniques using Chicago homicide data. While all three methods attempted to identify areas with elevated concentrations of crime, each approached the problem differently.

The Grid Overlay method divided the study area into uniform grid cells and identified the cells containing the highest number of incidents. This approach was straightforward and easy to understand, but it also highlighted one of its limitations: the location of the grid boundaries can influence the results. Even small changes in grid placement could potentially change which locations are identified as hotspots.

---

The Kernel Density hotspot method produced a smooth surface that highlighted general areas of concentrated homicides rather than individual hotspot polygons. During this portion of the lab, I encountered a few challenges while converting the raster output into polygons and distinguishing between raster and vector datasets. Working through these issues improved my understanding of raster workflows and reinforced the importance of carefully managing intermediate outputs during spatial analysis.

---

The Local Moran's I analysis was particularly interesting because it identified statistically significant High-High clusters instead of simply locating areas with many incidents. After selecting only the High-High clusters and dissolving the results into a single hotspot feature, I compared this method with the Grid Overlay and Kernel Density approaches. Although all three methods identified a similar proportion of the 2018 homicides, the hotspot areas differed considerably.

---

To compare the effectiveness of each hotspot technique, I calculated the total hotspot area, determined how many 2018 homicides occurred within each hotspot, calculated the percentage of all homicides captured, and computed homicide density. The Grid Overlay method produced the smallest hotspot area and the highest homicide density, while the Kernel Density method identified a larger hotspot area with a lower density. Local Moran's I produced results that balanced hotspot size with statistically significant clustering.

From a practical standpoint, I found Local Moran's I to be the most useful method for supporting law enforcement decision-making. Rather than simply identifying areas with many incidents, it identifies clusters that are statistically significant, making it a stronger analytical tool for directing limited enforcement resources. While Kernel Density remains an excellent visualization technique and Grid Overlay provides a simple summary of crime concentration, Local Moran's I offers the strongest foundation for strategic planning because it accounts for the spatial relationship between neighboring locations.

Overall, this lab strengthened both my technical skills and my understanding of spatial crime analysis. In addition to practicing raster analysis, polygon processing, spatial selection, and geometry calculations, I gained a better appreciation for how analytical methods influence the interpretation of crime patterns. It also reinforced that effective GIS work involves more than running tools—it requires understanding what each method measures, recognizing its limitations, and selecting the approach that best answers the question being asked.