Abstract
This study examines the spatial distribution of Airbnb listings in Albany, New York, and evaluates their statistical clustering using the Getis-Ord Gi* local indicator of spatial association. Listing locations were aggregated into 8,000-foot hexagonal areal units to address spatial structure and mitigate point-level noise. A fixed distance band of 12,000 feet with row-standardized spatial weights was applied to identify statistically significant hot and cold spots of short-term rental concentration. The resulting spatial patterns were compared with FEMA-designated flood hazard zones to assess potential spatial overlap between economic clustering and environmental exposure. Results indicate statistically significant concentration of listings in central and river-adjacent areas, with partial overlap observed in mapped flood hazard corridors. The study demonstrates the importance of spatial aggregation, scale calibration, and methodological transparency in urban spatial analysis.
1. Introduction
The growth of short-term rental platforms has generated increasing interest in understanding their spatial distribution within urban environments. While descriptive mapping of listing locations provides visual insight, it does not distinguish between random spatial variation and statistically significant clustering.
This study applies spatial statistical methods to identify localized clustering of Airbnb listings within the City of Albany, New York. Specifically, the objectives are:
- To determine whether Airbnb listings exhibit statistically significant spatial clustering.
- To evaluate the spatial relationship between identified clusters and FEMA-designated flood hazard zones.
By integrating local spatial autocorrelation analysis with environmental risk mapping, the study aims to provide a structured spatial assessment of short-term rental concentration and potential exposure.
2. Study Area and Data
2.1 Study Area
The study area consists of the municipal boundary of Albany, New York. All spatial analysis was constrained to the city boundary to ensure consistency in spatial weights and aggregation.
2.2 Data Sources
The primary dataset consisted of geocoded Airbnb listing locations obtained from the Inside Airbnb platform. Each record represents an active listing within the study area.
Flood hazard information was obtained from the FEMA National Flood Hazard Layer (NFHL), which provides polygon delineations of flood risk zones.
All datasets were projected to NAD 1983 StatePlane New York East (US Feet) to support distance-based spatial analysis.
3. Methodology
3.1 Spatial Aggregation
Point-level Airbnb listings were aggregated into hexagonal tessellation units. A hexagonal grid was generated across the study area using an 8,000-foot cell size.
Hexagons were selected because they:
- Provide uniform area units,
- Maintain consistent centroid-to-centroid distances,
- Reduce directional bias associated with square grids,
- Support stable neighborhood structures in local spatial statistics.
Each hexagon was assigned a count of listings using spatial join operations.
The 8,000-foot cell size was selected following preliminary testing of smaller units, which produced excessive fragmentation and reduced interpretability at map layout scale. The chosen scale balances analytical resolution with spatial stability.
3.2 Hot Spot Analysis
Spatial clustering was evaluated using the Getis-Ord Gi* statistic. The Gi* statistic measures the degree to which high or low values cluster spatially by comparing local sums to the global mean within a defined neighborhood.
Parameters used:
- Analysis variable: Listing count per hexagon
- Conceptualization of spatial relationships: Fixed Distance Band
- Distance band: 12,000 feet
- Spatial weights: Row-standardized
The 12,000-foot distance band approximates 1.5 times the hexagon size, ensuring sufficient neighborhood connectivity while preserving localized clustering patterns. Row standardization was applied to normalize weight influence across hexagons with varying numbers of neighbors, particularly near study area boundaries.
The output classification was based on Gi* z-scores and associated p-values, identifying:
- 99%, 95%, and 90% confidence hot spots,
- Statistically non-significant areas,
- Cold spots.
3.3 Flood Hazard Overlay
FEMA flood hazard polygons were overlaid on the Gi* classification output in a separate panel to assess spatial correspondence between statistically significant Airbnb clusters and mapped flood exposure zones.
Both map panels were constrained to identical scale and extent to ensure valid comparative interpretation.
4. Results
The Gi* analysis identified statistically significant clustering of Airbnb listings in central Albany and river-adjacent areas. These hot spots represent areas where listing counts are significantly higher than expected under spatial randomness.
Cold spots were observed in peripheral and lower-density zones, indicating areas with significantly lower listing concentrations.
Overlay analysis revealed partial spatial overlap between certain statistically significant clusters and FEMA-designated flood hazard corridors. While the study does not infer causal relationships, the spatial coincidence suggests areas where concentrated short-term rental activity intersects with mapped environmental exposure.
5. Discussion
The results demonstrate that Airbnb listings in Albany are not randomly distributed but exhibit measurable spatial clustering. The application of local spatial autocorrelation provides a more defensible assessment than descriptive point density mapping.
The analysis also highlights the influence of aggregation scale, reflecting considerations related to the Modifiable Areal Unit Problem (MAUP). Initial tests using smaller hexagon sizes increased spatial noise and reduced interpretability, underscoring the importance of scale calibration in local statistical analysis.
The integration of environmental risk layers provides contextual insight into potential exposure patterns. While overlap does not imply vulnerability or hazard realization, it provides a spatial basis for further investigation.
6. Conclusion
This study applies hexagonal aggregation and Getis-Ord Gi* analysis to identify statistically significant spatial clustering of Airbnb listings in Albany, New York. The findings reveal concentrated short-term rental activity in central and river-adjacent areas, with partial overlap in mapped flood hazard zones.
The project illustrates the importance of:
- Thoughtful spatial aggregation,
- Calibrated distance-based weighting,
- Transparent methodological documentation,
- Cartographic refinement aligned with analytical clarity.
By combining spatial statistics and environmental context, the analysis moves beyond visualization toward structured spatial inference.