Datahack @ Yale 2017

This code was developed in 2 days at Datahack@Yale 2017 and won 2nd place overall out of 15 teams.

Yale Policy Lab Challenge

The challenge from the Yale Policy Lab and the Justice Collaboratory was to better understand police misconduct by analyzing public data on pollice officer complaints from the Chicago Police Department collected by the invisible institute.

Part of the challenge was to use statistical modeling to predict future police misconduct while the other part was effectively visualizing the network of complaints to aid in interpretation.

Data Visualization

We visualized the co-complaint network where each officer is a node and they are linked if they were implicated in the same complaint (click here to view in browser).

Data dictionary

data/toy.officer_data_cleaned.csv

• officer_id
• first_name
• last_name
• appointed_date
• race
• gender
• birth_year
• age
• rank
• primary
• secondary
• tertiary

data/toy.complaint_data_cleaned.csv

• crid - complaint record ID
• officer_id - the officer ID
• incident_date - date of incident
• incident_time - time of incident (00:00 to 23:59)
• district - the district extracted from beat_2012_geocoded
• beat_2012_geocoded - beats
  • A police beat is an area.
  • Described by integer up to 4 digits.
  • A beat is a subset of a district.
  • The first two digits of a beat represent the District.
• complaint_category - category of the complaint
• complaint_name - category of the complaint
• final_finding - The final finding
  • EX: Exonerated
  • UN: Unfounded
  • NAF: No Affidavit
  • NS: Not Sustained
  • SU: Sustained
  • NA: Unknown