Search for a command to run...
This study presents a machine learning and statistical framework for analysing seasonal temperature trends across the United Kingdom and generating a 10-year forecast of future temperature patterns. Using the HadUK-Grid gridded climate dataset provided by the UK Met Office, the study examines historical temperature observations between 2010 and 2023 across England, Scotland, and Wales. The objective is to identify seasonal warming patterns and evaluate the potential of machine learning models to support medium-term climate forecasting. Multiple modelling approaches are implemented, including classical statistical regression and machine learning algorithms such as Gradient Boosting and Random Forest. Time-series modelling techniques, including SARIMAX, are used to capture seasonal dynamics and temporal dependencies within the climate data. Model performance is evaluated using standard statistical metrics including Root Mean Square Error (RMSE) and coefficient of determination (R²), allowing comparison between predictive methods. Results indicate a statistically significant warming trend in summer temperatures across all three UK nations, with projections suggesting continued seasonal temperature increases through the forecast horizon of 2026–2035. The analysis highlights the potential role of machine learning in climate monitoring and medium-term forecasting, providing insights that may support climate risk assessment, environmental planning, and infrastructure resilience strategies. The research demonstrates how open climate datasets combined with interpretable machine learning techniques can produce scalable analytical frameworks for national-scale climate analysis. The approach presented in this study can be adapted to other regions and extended to additional climate variables such as precipitation, heat stress indices, and extreme temperature events. All climate data used in this study are publicly available from the HadUK-Grid dataset provided by the UK Met Office and distributed through the Centre for Environmental Data Analysis (CEDA). The modelling pipeline was implemented in Python using libraries including pandas, scikit-learn, and statsmodels.