4 ms·
$ curl https://raw.githubusercontent.com/washingtonpost/data-homicides/master/homicide-data.csv https://raw.githubusercontent.com/washingtonpost/data-homici...
by a_bonobo 8y ago
$ curl https://raw.githubusercontent.com/washingtonpost/data-homicides/master/homicide-data.csv https://raw.githubusercontent.com/washingtonpost/data-homici... | grep 'Closed by arrest' | wc -l
25667
$ curl https://raw.githubusercontent.com/washingtonpost/data-homicides/master/homicide-data.csv https://raw.githubusercontent.com/washingtonpost/data-homici... | grep -v 'Closed by arrest' | wc -l
26506
Wow that's depressing. All possible outcomes via
$ curl https://raw.githubusercontent.com/washingtonpost/data-homicides/master/homicide-data.csv https://raw.githubusercontent.com/washingtonpost/data-homici... | cut -f 12 -d ',' | sort | uniq -c
1 Closed by arrest
25673 Closed by arrest
2922 Closed without arrest
1 disposition
23583 Open/No arrest
Pretty sure that 'Closed by arrest' also regularly results in a 'Not guilty' down the pipeline, so the 'real' number is probably much lower.
- Pywarrior 8y agoIt's Closed without arrest and Open/No Arrest that are most concerning. Closed by arrest indicates effort was probably spent during an investigation, the bare minimum assumption. An assumption that can not be made about the 26,505 Closed without arrest and Open/No Arrest murders.