Introduction
This report investigates crime patterns in Liverpool using geographic data and deprivation indicators. It aims to visualize crime distribution and develop a machine-learning classifier to predict crime levels based on socioeconomic factors.
Aims
The study focuses on visualizing crime distribution, identifying crime hotspots, and using deprivation data to classify crime severity levels. It seeks to understand the correlation between population density and crime occurrence.
Methods
The report utilizes thematic maps and cartograms to visualize crime data, while decision tree and random forest algorithms classify crime levels. Data from UK national statistics and Merseyside Police are used, covering November 2019. The accuracy of the classification models is compared.
Results
Crime Distribution: The highest concentration of crimes occurs in the port area and city center. The most common crime types are violence, antisocial behavior, and criminal damage.
Correlation Analysis: A weak correlation is found between population and crime rates.
Classification Models: The decision tree model achieves 60.34% accuracy, while the random forest model reaches 62.07%.
Conclusion
The findings highlight crime hotspots in Liverpool and the limited predictive power of deprivation indicators for crime classification. Model accuracy may improve with larger datasets and additional features.
Limitations
The study is limited by the small dataset, lack of outlier detection, and minimal exploration of alternative models. Future work will address these issues to enhance model accuracy.