Data and code for Sensitivity of pedestrian rhythm indicators to clustering resolution and count scale

This record contains the data, analysis code and computational outputs accompanying the manuscript “Sensitivity of pedestrian rhythm indicators to clustering resolution and count scale.” The study examines temporal breadth, day type diversity and rhythmic stability using the MONSTER Pedestrian dataset, derived from the City of Melbourne Pedestrian Counting System. The source archive contains 189,621 daily profiles of 24 hourly pedestrian counts across 82 location classes, with documented coverage between 2009 and 2022. Excluding 133 all-zero profiles leaves 189,488 profiles for distributional analysis. The package includes the unmodified source count and label arrays; source documentation and checksums; Python scripts for indicator calculation, clustering, resampling, simulation, regression and figure generation; machine-readable derived results; software requirements; and instructions for reproducing the analyses. The workflow was reconstructed for the revised manuscript, with explicit analytical settings and fixed random seeds. The analyses assess sensitivity to clustering resolution, initialisation, unequal numbers of daily records and count scale. Bootstrap and random-half comparisons assess internal repeatability. Multinomial simulations examine whether associations between pedestrian volume and estimated stability can arise from count scale alone. These simulations are diagnostic experiments rather than corrected observations or evidence of behavioural mechanisms. The distributed source files do not provide calendar dates or a verified mapping to named sensor locations. Consequently, the results do not establish land-use explanations, social diversity, public-space quality or causal effects of design interventions. Internal repeatability should not be interpreted as chronological or external validation. Original data and source documentation retain their CC BY 4.0 licence and attribution to the City of Melbourne and MONSTER contributors. The revision analysis scripts are provided under the MIT licence. The included source-repository loader retains its Apache 2.0 licence. Source dataset: MONSTER Pedestrian dataset.