How RentSift reads Montgomery County crime and complaint data

MoCo crime data lives at data.montgomerycountymd.gov under dataset ID icn6-v9z3. Each record includes a start_date, county crime category fields such as crimename1 and crimename2, and latitude/longitude coordinates.

We query using Socrata's within_circle geo filter, passing the address coordinates and a 0.5-mile radius (about 805 meters). This is the same radius we use for DC crime data, which keeps the comparison roughly consistent across jurisdictions.

The query also filters by date, pulling only the last 12 months of incidents. Older data is still in the dataset but falls outside the window that a renter cares about when evaluating a current address.

The current loader asks the MoCo API for records where crimename1 is "Crime Against Person" or "Crime Against Property." That avoids guessing from free text, but it also means the report is intentionally focused on those categories instead of forcing every police record into a renter-facing bucket.

DC and MoCo still use different source schemas and offense labels. A DC OFFENSE value and a MoCo crimename2 value are not always one-to-one, so cross-county comparisons should focus on radius, date window, and incident mix rather than a single merged verdict.

Records outside the person/property categories may still matter in real life, but they are not treated as violent or property incidents unless the source category supports that mapping.

MoCo has a 311 service request dataset (jrcn-in39 on the same Socrata portal), and we do pull from it. Because that dataset is not a clean address-radius feed for our use case, RentSift uses a ZIP-first and city fallback rather than treating it as exact building evidence.

For stronger property-condition context, the report also uses Montgomery County Housing Code Violations when the source supports a geospatial radius query. Those records can be separated into at-building versus nearby context when coordinates are available.

The complaint categories in MoCo's systems do not map directly to DC's SERVICECODEDESCRIPTION values. Until each county has its own tested mapping, the report keeps source notes visible instead of pretending all complaint feeds are equivalent.

MoCo reports include nearby police-report counts and complaint context with source notes. When the classification is partial, the copy says so instead of turning incomplete mapping into a confident-looking grade.

The most useful renter question is not whether a county-wide model likes the address. It is whether the records near this pin show a pattern worth asking the landlord about, and whether the report is talking about this property or nearby public data.

The next concrete step is continuing to test county-specific mappings and source labels against real addresses. Until then, the report keeps the source and limits visible.