Predictive analytics relies heavily on time series forecasting, which influences choices in a variety of fields. Using statistical, hybrid machine learning, and deep learning techniques, this study assesses many forecasting models on four datasets. Stationarity tests, automation with Sweetviz for exploratory analysis, and data pretreatment are all part of the methodology. Before using different models, such as ARIMA, SARIMA, and their hybrid integrations with Random Forest, XGBoost, LightGBM, and AdaBoost, STL decomposition is used to extract seasonality, trend, and residual components. We also evaluate FB Prophet, a model that is intended to successfully manage trend fluctuations and seasonality. Time series forecasting is essential to predictive analytics and influences choices in many different fields. Using deep learning, hybrid machine learning, and statistical methods, this study assesses many forecasting models on four datasets. Data pretreatment, stationarity tests, and automation with Sweetviz for exploratory analysis are all part of the technique. ARIMA, SARIMA, and their hybrid integrations with Random Forest, XGBoost, LightGBM, and AdaBoost are among the models that are implemented when STL decomposition is used to extract seasonality, trend, and residual components. We also evaluate a model called FB Prophet, which is intended to successfully manage seasonality and trend fluctuations.