Hyperparameter Tuning for Identifying Tree Seedling Mortality Using Logistic Regression on Environmental and Factors

Tree-seedling mortality is a significant challenge in forest management and agricultural practices, particularly in areas affected by climate change. The primary causes of seedling mortality are related to environmental factors and soil conditions. The early identification of these factors is crucial for improving seedling survival and land productivity. While existing studies have applied machine learning techniques to predict seedling mortality, they have often overlooked the importance of hyperparameter tuning in enhancing model performance. This study addresses this gap by focusing on the impact of hyperparameter tuning on the Logistic Regression models used to predict tree seedling mortality. Using a tree survival prediction dataset containing diverse environmental and soil factors, the model was optimized to achieve significant improvements. The tuned Logistic Regression model outperformed the standard approaches, achieving an accuracy of 93.71% and AUC of 97.81%. These results underscore the importance of hyperparameter optimization in boosting predictive accuracy, and demonstrate its potential to support improved forest conservation and management efforts more effectively than traditional models.