This vignette introduces the `forecast` package for R, which implements automatic forecasting algorithms for univariate time series. The package includes two main algorithms: one based on innovations state space models underlying exponential smoothing methods, and another based on a step-wise algorithm for ARIMA models. Both algorithms are applicable to both seasonal and non-seasonal data. The vignette also describes the implementation of these methods in the `forecast` package, along with other features such as prediction intervals and model selection criteria. The package is demonstrated using four real-time series examples, and the results are compared to those from other forecasting methods. Additionally, the vignette provides an overview of the `forecast` package's functionality, including functions for exponential smoothing, ARIMA modeling, and other forecasting techniques.This vignette introduces the `forecast` package for R, which implements automatic forecasting algorithms for univariate time series. The package includes two main algorithms: one based on innovations state space models underlying exponential smoothing methods, and another based on a step-wise algorithm for ARIMA models. Both algorithms are applicable to both seasonal and non-seasonal data. The vignette also describes the implementation of these methods in the `forecast` package, along with other features such as prediction intervals and model selection criteria. The package is demonstrated using four real-time series examples, and the results are compared to those from other forecasting methods. Additionally, the vignette provides an overview of the `forecast` package's functionality, including functions for exponential smoothing, ARIMA modeling, and other forecasting techniques.