3.3 Moran's I

Moran's I is a measure of spatial autocorrelation, which quantifies how similar or dissimilar data values are based on their spatial location. This analysis is particularly useful for understanding the clustering patterns within spatial data. Our analysis focuses on identifying spatial clusters and hotspots during various time intervals. The analysis employs both global and local Moran's I to assess spatial patterns.

Global Moran's I Values

Global Moran's I values indicate the degree of spatial autocorrelation within the data, highlighting patterns of clustering or dispersion. The analysis reveals a noticeable trend: the Moran's I values are higher during the pre-Corona period and the phases of relaxing restrictions. This indicates strong spatial clustering, where similar values (e.g., high activity levels) are concentrated in neighboring areas.

Local Moran's I Values

For the Local Moran's I the p-value is incorporated to filter significant results, ensuring only relevant spatial clusters are highlighted. Therefore, only points with a p-value below 0.05 are displayed. This approach not only reduces unnecessary data but also optimizes the performance of this web-based visualization tool. By focusing on significant clusters, we aim to provide clear and actionable insights without overloading the system.

Pre-Corona

Pre-Corona period shows clear spatial clusters with positive Moran's I values. Negative values cluster around roads, highlighting differences between roads and surrounding areas. Polygons of Interest (POIs) like bus depots and subway stations frequently appear in red areas, indicating contrasting patterns compared to surrounding regions.



Results and Interpretation

The Moran's I analysis, incorporating both global and local perspectives, highlights dynamic spatial patterns and their evolution across different time intervals during the pandemic. These patterns reveal significant insights into the clustering of activities and variations in key areas such as roads and Points of Interest (POIs).

Global Moran's I values consistently show higher spatial autocorrelation during the Pre-Corona period and relaxation phases, indicating strong clustering of activities in urban centers and hotspots. Conversely, during lockdown phases, the values decline, reflecting a reduction in clustering and the homogenization of spatial patterns as mobility decreased.

Local Moran's I reveals finer details, highlighting specific areas where spatial clusters (hotspots and cold spots) occur. Negative Moran's I values, often observed around roads, suggest contrasting activity patterns between roads and surrounding areas. POIs such as bus depots and subway stations frequently appear in these negative-value zones, particularly during the Pre-Corona and Relaxation phases, underlining their distinct roles in the spatial activity landscape.

The following key observations stand out:

This comprehensive analysis demonstrates the utility of Moran's I in uncovering spatial dynamics over time. By integrating both global and local perspectives, it becomes possible to pinpoint significant variations and contrasts, offering actionable insights for understanding spatial activity patterns.