Police misconduct worldwide is a major problem and involves incidents like fatal shootings, mistreating citizens and involvement in corruption activities. Various preventive measurements have been taken up to date to tackle these issues like fitting police cars with recording systems and introducing penalties for misconduct but still such cases exist and are frequently reported. Algorithms also exist to predict bad cop behavior. Many departments have programs that based on history and patterns predict when and which cop will most likely be involved in misconduct so that their supervisors can beforehand send them to counseling and keep them under surveillance.
Researchers from University of Chicago are now working to make a better early warning system. Researchers have background in statistics, programming, economics and related disciplines. To develop their algorithm they analyzed over a decade of data from Charlotte police of cops and public interaction. They found out that the most prominent factor in determining bad conduct of an officer is his previous history. An officer having three to four misconduct reports in an year is likely to repeat the same behavior in the next year. Using this and many other indications the Chicago University’s algorithm was better able to predict trouble.
The idea of using past history to predict bad cop behavior is not new and has been used before. Chicago Police Department had also developed such algorithm 20 years ago but had to abandon it due to officer’s resistance. The software used neural-network-based data analysis to predict bad police conduct and immediately gained a lot of media attention. The software was however rejected by officers since they didn’t like to be apprehended for something they hadn’t committed.
Compared to older algorithms this new algorithm predicts 51 percent fewer officers and also has produces more accurate results.
The Wonderful World of Algorithms 