| This study investigates the impact of climatic factors on the growth of cork oak (Quercus suber) in Tunisia, applying machine learning techniques, specifically regression and classification models, enhanced by cascading classification with clustering. Initial analysis showed weak direct correlations between growth and individual climate variables; however, notable inter-variable correlations, such as a 0.74 link between winter and spring maximum temperatures, highlighted significant indirect influences. The Random Forest regression model achieved a robust performance with an accuracy of 83% and Mean Squared Error (MSE) of 0.02, effectively modeling complex climate-growth relationships. Additionally, incorporating K-means clustering into the Decision Tree classifier increased accuracy from 64% to 76%, showcasing the adaptability of cascading classification for heterogeneous environmental data. This machine learning approach enables the extraction of interpretable rules, identifying summer rainfall and spring temperature as key growth drivers. These findings demonstrate the utility of advanced data science techniques in forest management, providing actionable insights into adaptive strategies for ecosystems under climate variability. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.