A time-series model describes data indexed by time, capturing temporal dependencies, trends, seasonality, and irregular fluctuations to explain historical patterns or forecast future values. section 7.5
Common approaches differ in the temporal structure they model:
- ARIMA combines autoregressive, differencing, and moving-average components; differencing transforms nonstationary data into a stationary series. section 7.5
- Simple exponential smoothing weights recent observations more heavily and suits series without trends or seasonal patterns. Double exponential smoothing adds a trend component. section 7.5
Forecasting applications
Time-series models can support forecasting stock prices, retail demand, transportation traffic, and energy consumption. section 7.2
