The survival of forest ecosystems relies heavily on the health of trees. Therefore, it is essential to utilize data- driven analytical tools for predicting the probability of survival, which can inform forest management de cisions. In this research, machine learning classification algorithms were employed to predict the survival of Carpinus betulus trees in the forests of Northern Iran. Several models were trained using long-term data from permanent ground sample plots, including Support Vector Machine (SVM), Random Forest (RF), and LightGBM. Key variables in the analysis included the diameter at breast height (DBH) and the basal area of the largest trees (BAL). The results demonstrated that the LightGBM model achieved the highest performance with respect to the balanced-accuracy measure. These findings provide valuable insights for developing effective forest conservation and management strategies. The study highlights the capability of machine learning algorithms to accurately predict tree survival, thereby contributing to improved forest management and conservation efforts in dynamic ecosystems.
