Streamline cops filter - Githubissues

openjusticebaltimore / gttf

Investigation of the prevalence of Baltimore Police Department's Gun Trace Task Force testimony in State of Maryland judicial cases

4 stars 2 forks source link

Streamline cops filter #2

Open kbmorales opened 4 years ago

kbmorales commented 4 years ago

Currently using string detection, but could use something more concrete like officer ID

kbmorales commented 4 years ago

@camille-s has done some work on this!

kbmorales commented 4 years ago

Note: common misspelling of Hankard is Hanford

camille-s commented 4 years ago

Second pass to improve upon #3:

[x] More string cleanup
[ ] ~~Pull from dsk8 related persons, not just dscr—probably roughly the same cops in both, but might as well be more exhaustive~~ Ooof, officer IDs in dsk8 are all null
[x] Incorporate manual running list of corrections (e.g. Hankard)
[ ] Concatenate names that look like maiden name --> married name: same ID & first name, 2 last names used

camille-s commented 4 years ago

Manual name correction

Manual abbreviation corrections

camille-s commented 4 years ago

Starting on a second script now to clean up names from dsk8. There are no officer IDs in dsk8 (wtf), so I'm going to clean their names, then fuzzy-match to the cleaned-up names this script generates. I'm assuming all or most officers that have been on a case in circuit court (dsk8) have also been in district court (dscr). This will let us do some of the same analysis of charges, nolles, networks, etc on circuit court cases that we're doing on district using IDs instead of messy strings.