HomeRiskIQ private review evidence package

Run python analyze.py using Python 3 standard library. Decimal comparisons enforce strict greater-than without binary floating-point boundary errors. Results: summary.csv, sensitivity.csv, runs.csv and monthly.csv. charts.py requires matplotlib and regenerates the comparison and timeline images. build_doc.py requires python-docx and those images. The original JSON is in data; request URLs and retrieval date are in sources.json; checksums.sha256 fixes the evidence snapshot.

Counts are sampled gauge readings, not continuous flood duration or property impacts. Runs cross midnight and break at non-exceedance, flagged data or a missing/nonconsecutive hour. UTC day aggregation is explicit. See the review document for complete methods and limitations.
