The data behind this site, ready to download

Start here

• /data/index.json — the machine-readable catalog of everything below (URLs, schemas, sizes).
• /llms.txt — the same map, for AI assistants and agents.
• /data/manifest.json — bundle version 4.0.0 and per-source as-of vintages.

Join key everywhere: 3-digit MCPS school code, zero-padded string (e.g. 427 = Walt Whitman HS). Suppression: null values alongside suppressed flags are privacy-suppressed small counts — do not impute or infer them.

Where to look

The question-to-file map, straight from the catalog — the same routing served to AI agents in /llms.txt.

Datasets

Geo files

• /boundaries/ES.geojson
• /boundaries/MS.geojson
• /boundaries/HS.geojson
• /boundaries/MS_adopted.geojson
• /boundaries/HS_adopted.geojson
• /bus/{code}.geojson — one GeoJSON per school (3-digit code)

Query the fact tables directly

These Parquet tables carry every figure behind this site's pages, plus many breakdowns no page shows: MCAP and SAT/ACT/AP results, accountability, graduation, climate, finance, educators, enrollment, capacity, and neighborhood profiles. The full list, with URLs and field docs, is in the table above and in index.json. Start with /data/school_year.parquet — one row per school and year with the headline measure from every source already joined (a ~230 KB CSV twin needs no Parquet reader at all). The full-fidelity archive,/data/mcps_consolidated.parquet, holds every source cell exactly as published (~3M rows) — use it to verify a number, not to start from. Any Parquet-aware tool can query these over HTTP without downloading them first — for example, inDuckDB:

SELECT school_name, accel_pct, math_all_pct, farms_pct
FROM 'https://mocoparents.org/data/school_year.parquet'
WHERE school_type = 'MS' AND year = 2025
ORDER BY accel_pct DESC LIMIT 5;

Column meanings:mcps_consolidated_columns.json(consolidated), mcps_rows.schema.json, and msde_dictionary.json. Null values withsuppressed = true (MCPS) or marker strings in value_text (MSDE) are privacy suppression by design. Where both sources report the same fact, useWHERE canonical for one answer per fact (enrollment prefers MCPS; MSDE-origin rates and scores prefer MSDE), and treat discrepant = true rows with care — the sources genuinely disagree there (cross-check report).

Charts straight from SQL

The tables above aren't just downloads — the site itself builds charts from them. This one is a DuckDB query against mcps_consolidated.parquet executed at build time and shipped as inline SVG (no JavaScript): countywide enrollment from the canonical cross-source rows, where MCPS is preferred when both systems report a grade and MSDE fills the earlier years.

SELECT year::INT AS year, SUM(value)::DOUBLE AS students
FROM mcps_consolidated
WHERE concept = 'enrollment_by_grade_count' AND canonical
GROUP BY year ORDER BY year

Terms

All data derives from public MCPS, MSDE, and U.S. Census sources; this site's transformations are provided as-is, for any use, with attribution appreciated ("MoCo Parents, mocoparents.org"). Please don't present suppressed values as zeros, and check each dataset's schema before comparing numbers across sources.