Spatial Analysis of Violent Crime Dataset Using Machine Learning
No Thumbnail Available
Date
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Description
The monster called crime has been living with us from the beginning of human
existence and impacts negatively on the general health of a nation. Different
approaches were employed in the past studies for predicting occurrence of
violent crime to aid predictive policing, which makes conventional policing
more efficient and proactive. This paper investigates the accuracy of Machine
Learning-based crime prediction approaches, which were used previously by
other researchers. This study presents Machine Learning approaches to
violent crime prediction. Five years’ historical dataset between July 2014 and
July 2019 were collected from Nigerian Police Lagos, analyzed and used for
training the models built. Two different Machine Learning predictive models,
Decision Tree and K-Nearest Neighbor, were implemented using IBM Watson
Studio and violent crime prediction accuracy of 79.65%, and 81.45% were
obtained, respectively, with the real-life dataset collected from Nigerian Police
Obalende Lagos and online crime reported portal during violent crime
prediction in Lagos. This could be used to enhance crime prevention and
control strategies in curbing the worrisome crime rate in the country.
Keywords
Q Science (General), QA76 Computer software