23 attribute fields, profiled from the April 2018 vintage.
Columns in Midland County
| Field | Type | Filled | Distinct | Example values |
|---|---|---|---|---|
| PARCEL_ID | OFTString | 40.7k | 170-007-100-210, 170-007-100-410, 170-007-100-200 | |
| SITUS | OFTString | 34.8k | 104 E ADAMS ST, 105 E ADAMS ST, 108 E ADAMS ST | |
| ADDR_NUMBE | OFTString | 6.1k | 104, 105, 108 | |
| STREET_PRE | OFTString | 4 | E, E, E | |
| STREET_NAM | OFTString | 1.3k | ADAMS, ADAMS, ADAMS | |
| STREET_SUF | OFTString | 3 | N, N, N | |
| STREET_TYP | OFTString | 21 | St, St, St | |
| SITUS_CITY | OFTString | 14 | COLEMAN, COLEMAN, COLEMAN | |
| SITUS_ZIP | OFTString | 16 | 48618, 48618, 48618 | |
| ADDR_SEC_U | OFTString | 2 | STE, SUITE, SUITE | |
| ADDR_SE_01 | OFTString | 4 | 1, A, A | |
| OWNER | OFTString | 32.5k | SIGSBY ANN L AND RICKY R W&H, MALTBY TY R, COUNSELMAN JUSTIN & AMBER | |
| SALE_PRICE | OFTReal | 2.1k | 0.0, 0.0, 0.0 | |
| LAND_USE_C | OFTString | 38 | 4011, 4011, 4000 | |
| LAND_US_01 | OFTString | 4 | Residential, Residential, Residential | |
| MUNI_ID | OFTInteger64 | 19 | 1626110, 1626110, 1626110 | |
| SCHOOL_DIS | OFTInteger64 | 11 | 2610200, 2610200, 2610200 | |
| MKT_VAL_LA | OFTReal | 421 | 0.0, 0.0, 0.0 | |
| ZIP_CODE | OFTString | 16 | 48618, 48618, 48618 | |
| YEAR_BUILT | OFTInteger64 | 134 | 0, 0, 0 | |
| LEGAL_DESC | OFTString | 16.2k | NW 1/4 OF NE 1/4 OF SEC 12, T14N-R1E, UNIT 6 THE COURTS OF JOSEPHS RUN CONDOMINIUM, UNIT 5 THE COURTS OF JOSEPHS RUN CONDOMINIUM | |
| MAIL_ADDRE | OFTString | 4.7k | COLEMAN MI 48618, COLEMAN MI 48618, COLEMAN MI 48618 | |
| MAIL_AD_01 | OFTString | 30.0k | 104 E ADAMS ST, 105 E ADAMS ST, 108 E ADAMS ST |
Why this differs from the next county over
There is no national parcel schema. Every assessor's office built its own, usually decades ago, usually inside whatever CAMA system it bought at the time. One county's owner column is OWNER_NAME, the next one's is own1, and a third splits it across four fields. Shapefile's ten-character DBF limit truncated a generation of them into things like TOTLNDVAL.
Nothing here is renamed or remapped. The columns above are exactly what is in the file, because a normalisation layer that quietly guesses wrong is worse than no normalisation at all — and every guess it makes is one you cannot audit.
The Filled column is the share of rows with a real value. Placeholder text such as n/a counts as empty; a column that is 100% present but 90% placeholder is a column that will disappoint you, and the profiler treats it accordingly.