This project will use natural language processing and inferential statistics to identify which public defenders, from case records, consistently (or at an abnormal frequency) seek plea bargains for their defendants. As well, it will use regression analyses to look for relationships between kinds of cases and defendants, assuming the demographic data can be cross-referenced from somewhere. This could make for a great partnership with the Innocence Project.
Some random feature ideas:
- In addition to the between-defender stats mentioned above, we can compare stats (e.g. guilty plea percentages) between public and private defenders. Such a publicized comparison could limit the potential for public defenders gaming our system via a consensus to uniformly seek plea bargains.
- Access to this data (case records) is highly siloed/compartmentalized and differs by county, state, and fed levels. Thus, it might be best to focus on a single city/county (San Francisco?).
If you’re interested, contact teammate Jude Calvillo.