Evaluating Bias in Big Data Using H3 Spatial Indexing

The H3 geospatial indexing system has made a profound impact on geospatial data analysis. Its novel methodology of converting geographic areas into hierarchical hexagonal grids has revolutionized the management of extensive spatial datasets and streamlined the integration of varied data sources. This approach has shown particular promise in handling complex economic datasets, such as web sources, which present unique challenges due to their size and diversity. However, our study highlights significant limitations of the H3 system in spatial econometrics, specifically in relation to the Modifiable Areal Unit Problem (MAUP). By conducting simulations, we evaluated how different aggregation levels in H3 affect the detection of spatial autocorrelation using Moran’s I estimators. The results reveal considerable difficulties in accurately identifying spatial autocorrelation, a key aspect in analyzing geographic and economic datasets. This paper highlights the need for ongoing research to refine spatial data analysis tools. Future efforts should prioritize using real-world data from geographic and complex economic analysis, to minimize biases in H3 aggregations.