The COVID-19 pandemic has highlighted the critical need for accurate forecasting methods to predictmortality rates and inform public health interventions. This research evaluates the comparative perfor- mance of tree-based forecasting models, specifically Gradient Boosting, Random Forest, and Linear Model Trees, against traditional statistical forecasting approaches like ARMA and Random Walk models. The study is conducted using two distinct programming environments, Python and ScalaTion, to examine their influence on forecasting accuracy. The analysis employs Symmetric Mean Absolute Percentage Error (sMAPE) as the primary metric to assess the forecasting models. Findings indicate that tree-based methods outperform traditional models in predictive accuracy across both platforms, demonstrating their robustness in handling complex data pat- terns. Moreover, the research confirms high consistency between Python and ScalaTion implementations, with minor variations attributed to platform-specific numerical and optimization differences. Additionally, this thesis explores the impact of rolling validation combined with train-test splits on model performance, revealing that regression tree models maintain superior accuracy across multiple forecasting horizons. This study contributes to the literature by providing a comprehensive evaluation of traditional and tree-based forecasting methods in varying programming environments, offering insights into their suitability for pandemic-related forecasting applications.