Running weekly burglary near‑repeat analysis in ArcGIS Pro 3.2 using Space Time Pattern Mining with 250 m bins and 7‑day steps, but the hot spots look smeared and unstable week to week. Does anyone have solid resources or sample workflows on picking bin/time parameters, correcting address clustering and seasonality, and automating output maps for 0700 roll call?
Your 7‑day steps are likely drifting — anchor the cube with a Time Step Reference at Monday 07:00 and set Neighborhood Time Step=2 to steady EHSA. > workflows on picking bin/time parameters, correcting address clustering and seasonality, and automating output maps — pick bin size from the median nearest‑neighbor distance so you get about 1–2 events/bin/week (250 m may be wide), run Collect Events to handle address piles, and schedule an arcpy.mp export at 06:55 (Introduction to arcpy.mp—ArcGIS Pro | Documentation). If downtown is dense, try 200 m or segment‑based cubes instead.