Datasets:

Nason
/

fjc-database

license: other
task_categories:
  - tabular-classification
  - tabular-regression
tags:
  - courts
  - federal-courts
  - litigation
  - criminal-justice
  - fjc
  - duckdb
  - datapond
pretty_name: FJC Integrated Database (federal civil, criminal, appellate cases)
size_categories:
  - 10M<n<100M

FJC Integrated Database (IDB)

Every civil case (SY1988-present), criminal defendant (FY1970-present) and appeal (1971-present) reported by the federal courts to the Administrative Office of the U.S. Courts, as published in the Federal Judicial Center's Integrated Database (updated quarterly). Column names are the IDB's; -8/-9 and 01/01/1900 sentinels are stored as NULL. 21,119,023 rows across 3 tables.

Table	Description	Row Count	Column Count
`civil`	Civil cases filed in U.S. district courts, statistical year 1988 to present: one row per case (parties, nature of suit, jurisdiction, filing/termination dates, disposition, judgment)	10,959,535	47
`criminal`	Criminal defendants in U.S. district courts, FY1970 to present: one row per defendant (filing offenses, disposition, sentence, dates)	7,743,962	176
`appellate`	Appeals filed in U.S. courts of appeals, 1971 to present: one row per appeal (origin, nature of suit/offense, disposition, dates, judges)	2,415,526	71

Query it remotely

INSTALL httpfs; LOAD httpfs;
ATTACH 'https://huggingface.co/datasets/Nason/fjc-database/resolve/main/fjc.duckdb' AS fjc (READ_ONLY);

SELECT EXTRACT(YEAR FROM FILEDATE) AS year, COUNT(*) AS civil_cases
FROM fjc.civil WHERE FILEDATE >= '2015-01-01'
GROUP BY 1 ORDER BY 1;

Or with the datapond packages: pip install datapond / pak::pak("datapond-db/datapond-r").

Build pipeline, codebooks and caveats: https://github.com/ian-nason/fjc-database Source: https://www.fjc.gov/research/idb (public domain).