UK Crime Prediction Project — Gabriel Pierre Huertas de Las Puertas

Coventry University

This repository contains the full submission for the 6001CMD Machine Learning coursework.
It includes the academic report, the Jupyter Notebook used for data processing and modelling,
and the required presentation video.

Project Structure

├── 6001CMD_Coursework Template_25-26s1(1).docx ├── ASSESMENT.IA.ipynb ├── README.md └── simplescreenrecorder54-2025-12-01_12.24.25.mp4

File Descriptions

6001CMD_Coursework Template_25-26s1(1).docx
Complete academic coursework report including methodology, data sources, modelling, results, and conclusions.
ASSESMENT.IA.ipynb
Jupyter Notebook containing all code for data cleaning, feature engineering, exploratory analysis, model training, evaluation, and visualisation.
simplescreenrecorder54-2025-12-01_12.24.25.mp4
Presentation video summarising the project workflow, modelling techniques, and final findings.

Dataset Availability

The original datasets used in this project cannot be uploaded due to storage limitations and licensing restrictions.
However, the official dataset sources are listed below.

Dataset Links

(Insert your links here)

UK Police Crime Data (API & Monthly CSV):
https://data.police.uk/data/
UK Census 2021 (ONS / Nomis):
https://www.ons.gov.uk/visualisations/censusareachanges/data/mapdatadownload.csv
LSOA Boundary Shapefiles:
https://geoportal.statistics.gov.uk/datasets/ons::lower-layer-super-output-areas-december-2021-boundaries-ew-bsc-v4-2/about

Project Summary

This project focuses on predicting monthly crime counts in UK LSOAs (Lower Layer Super Output Areas) from 2022 to 2025.
The analysis integrates:

UK Police crime event data
Socio-economic indicators from Census 2021
Geospatial LSOA boundaries

Multiple regression models were evaluated, including:

Ordinary Least Squares (OLS)
Random Forest Regressor
XGBoost Regressor

Chronologically consistent train/validation/test splits were implemented, including the optimal split ratio proposed by Joseph (2022).

Requirements & Execution

To run the notebook, install the following Python libraries:

pandas numpy scikit-learn xgboost matplotlib seaborn

Optional for high-performance processing:

modin ray

It provides an overview of the methodology, system design, modelling steps, and key outcomes.

Author

Gabriel Pierre Huertas de Las Puertas
Coventry University
6001CMD — Machine Learning Coursework

License

This project is submitted exclusively for academic purposes under Coventry University regulations.
External datasets must be accessed and used according to the legal terms of their providers.

huertasdeg/16862171-GH-s1

About

Resources

Stars

Watchers

Forks

Releases

Languages