ДОСЛІДЖЕННЯ МЕТОДІВ ПРОГНОЗУВАННЯ ЧАСОВИХ ФІНАНСОВИХ РЯДІВ

This study systematically compares various time series forecasting methods, including classical statistical approaches and modern machine learning algorithms. Evaluation metrics such as RMSE and MAE are employed to assess predictive accuracy. Using real financial data from companies like Apple, Google, and Microsoft, the study evaluates ARIMA, LSTM, and XGBoost models. Results indicate XGBoost as the most accurate method. The findings underscore the importance of selecting appropriate forecasting methods tailored to specific contexts. Further research is encouraged to advance time series forecasting techniques.