URBAN CRIME INCIDENT TYPE CLASSIFICATION USING SPATIO-TEMPORAL AND CONTEXTUAL ATTRIBUTES

Abstract

Urban crime is a major challenge for modern cities, affecting public safety, urban planning,
and law enforcement strategies. With the rapid growth of urban populations and increasing
data availability, data-driven approaches can help authorities better understand and predict
crime patterns. This project focuses on the classification of urban crime incident types using
spatio temporal and contextual attributes derived from historical crime datasets. The proposed
system analyzes crime data based on multiple features such as geographic location, time of
occurrence, date, and contextual information like neighborhood characteristics and
environmental factors. Machine learning techniques are applied to identify patterns and
relationships between these attributes and different crime categories.
By preprocessing and analyzing large volumes of crime data, the system builds a predictive
model capable of classifying incidents into specific crime types with improved accuracy. The
methodology involves data collection, preprocessing, feature extraction, and the application
of classification algorithms such as Decision Trees, Random Forest, or Support Vector
Machines. The model is trained and evaluated using standard performance metrics including
accuracy, precision, recall, and F1-score. Visualization tools are also used to analyze spatial
and temporal crime distributions. The results demonstrate that incorporating spatio-temporal
and contextual attributes significantly improves the accuracy of crime type classification.
This approach can assist law enforcement agencies in crime analysis, resource allocation, and
proactive policing strategies. Ultimately, the proposed system contributes to smarter urban
safety management by leveraging machine learning and data analytics techniques