Abstract—Selecting an appropriate spatial resolution for aggregating fine-grained geospatial point data remains a key challenge in spatial analysis, particularly when aiming to remain agnostic to the downstream task. Building upon the framework proposed by Ramos, this paper introduces a distribution-free robustness metric based on normalized entropy, designed to avoid parametric assumptions and capture the heterogeneity of spatial configurations. We also assess the impact of binning strategies on the entropy calculation, comparing the classical binning methods to a data-driven approach based on the frequency of count values. Additionally, we also extended the method to higher dimensions, in order to incorporate temporal information. Our experiments confirm that the proposed metric behaves as expected: rising with granularity and inversely correlated with spatial uniformity. These findings support the method’s applicability to a wide range of spatial point patterns and motivate further investigation into its role in mitigating the Modifiable Areal Unit Problem (MAUP).