Machine Learning Models for Climate Prediction and Adaptation Planning in Tunisia: A Methodological Approach

Climate change poses significant challenges to urban planning in Tunisia, necessitating precise climate predictions for effective adaptation strategies. The methodology employs Random Forest regression algorithm with cross-validation to forecast temperature changes, incorporating historical meteorological data from Tunisia's Meteorological Institute. Model performance is assessed using Mean Absolute Error (MAE) as a metric of predictive precision. Random Forest models exhibited an MAE of 1.2°C for temperature predictions over three years, indicating moderate accuracy in climate forecasting. The developed machine learning models provide valuable insights into future climate patterns, which can inform urban planning decisions and enhance resilience to climate impacts. Urban planners should integrate these predictive models into their adaptation strategies to prepare for anticipated environmental changes. Random Forest regression, Climate prediction, Urban planning, Machine Learning, Tunisia Model estimation used $\hat{\theta}=argmin_{\theta}\sum_i\ell(y_i,f_\theta(x_i))+\lambda\lVert\theta\rVert_2^2$, with performance evaluated using out-of-sample error.