To create a model that accurately predicts crime while investigating the possible correlation between the prediction, demographics and social services. The purpose for our model is to assist in crime reduction via effective resource reallocation.
With CRISP-DM as guideline for approach: using 2014-2019 crime data from Toronto Police Service, supplemented with social services and demographics data, perform unsupervised and supervised machine learning.
The Toronto Police Service through their Public Safety Data Portal is "intended to improve the understanding of policing, improve transparency and enhance confidence through the creation and use of open data for public safety in Toronto". https://data.torontopolice.on.ca/
Unsupervised learning includes K-Mean and Hierarchical clustering.
Supervised learning includes linear regression, logistic regression, graident boosting decision tree, K nearest neighbour classifier, Support Vector classification, decision tree classifier, Naïve bayes classifier, Random forest classifier, and neural networks.
In addition, made attempts for time series analysis and sentiment analysis (using Twitter API).
File | Note |
---|---|
Crime Rate Predictive Analysis - FINAL - Team 2 - Janurary 2021.pdf | Final report |
York project - Toronto Police - 2021.pptx | Powerpoint summary |
Final Project - Clustering - Team 2.ipynb | Unsupervised learning - Jupyter notebook |
Final project - Exploration and Supervised - Team 2.ipynb | Supervised learning - Jupyter notebook |
Mapping of CRC to TPS.csv | source file for mapping rec centers to Toronto neighbourhoods |
Masterdf_crime.csv | source file for crime data combining social services and demographics |
PopdenNeig.csv | source file for population density by neighbourhood |
'Final project - Exploration and Supervised - Team 2.ipynb' references files 'PopdenNeig.csv' and 'Mapping of CRC to TPS.csv' to be in same folder as the Jupyter notebook
Suzanne Douglas
Rachna Kumari
Herby Robinson
Pushpendra Sharma
Don Sohn