Identifying Recidivism for Indian Police using Machine Learning

Abstract- Recidivism, the rate of crime by repeat offender, is as high as 35.5% in some states of India. Modus Operandi (MO), a term used to define set of habits that an offender follows to commit crime, is key to linking any crime to a repeat offender. Crime analysts spend countless hours going through data to determine if a crime fits into a known pattern. Since 2009, Indian Police has embarked on an ambitious program of computerizing more than 17000 police stations across India under the CCTNS project. CCTNS data contains Date, Time, Location, Suspect Details, Type of Crime, Victim Details, Details of Place of Occurrence, Properties seized, Arrested Person Details among other details related to crime. This research involves study of Modus Operandi of known criminal gangs and associated First Information Reports (FIRs) in the state of Chhattisgarh. Neural Network, Logistic Regression Gradient Descent and SVM Machine Learning algorithms were trained using Crime Data of known gangs to build classification models.