This project is a component of a larger project led by West Yorkshire Police to pilot predictive policing, funded by the Home Office through the Police Innovation Fund.
The main component of the project undertaken by the Quantative Criminology Group members runs for a year from February 2017 through until the end of February 2018. It is designed to test the effectiveness of the algorithm chosen for the project against alternative algorithms. Although the final output for this component of the project is a report detailing this comparison, this component will also develop a standardised testbed of predictive algorithms on a uniform codebed, along with a benchmark of statistical treatments for testing the algorithms, making a long term contribution to the assessment of predictive methodologies beyond the timeline of the project and enabling forces to choose algorithms suited to their specific circumstances. The various predictive implementations will be open source and freely available where licencing allows.
For details of the project as it progresses, see our GitHub repo. Note that the format for the scientific work in progress is Jupyter Notebooks, while code is in Python. Note that all work is in progress and should not be expected to work or represent solid research findings until more formally published.