DataTerminology.comNEW
Also known as: Seasonal ARIMA
An ARIMA model extended with seasonal differencing and seasonal AR and MA terms to capture periodic patterns.
DESCRIPTION
SARIMA generalizes ARIMA by including seasonal autoregressive and moving-average components plus seasonal differencing. It models both short-term dynamics and recurring seasonal patterns in a unified linear framework. Identification commonly uses ACF/PACF and seasonal diagnostics. Parameters are summarized by (p,d,q)×(P,D,Q)s with seasonal period s. SARIMA supports forecasting with confidence intervals and is widely available in statistical packages. With exogenous regressors, the model is often called SARIMAX.
ALTERNATIVE DEFINITIONS
Forecasting: Principles and Practice
A seasonal ARIMA model is formed by including additional seasonal terms in the ARIMA model.
Clear linkage to ARIMA via added seasonal terms.
Penn State STAT 510
A seasonal ARIMA model incorporates both non-seasonal and seasonal factors in a multiplicative model.
Highlights multiplicative structure of seasonal and non-seasonal parts.
statsmodels
Seasonal ARIMA (SARIMA) extends ARIMA by adding seasonal terms and seasonal differencing to account for periodic behavior.
Mentions seasonal differencing and periodic behavior explicitly.
RELATED TERMS
ARIMA Autoregressive Moving Average Seasonal Differencing Autocorrelation Function Partial Autocorrelation Box–Jenkins SARIMAX Stationarity Seasonality
COMPETING TERMS
Exponential Moving Average Prophet
KNOWLEDGE AREAS
TARGET AUDIENCE
Data Scientist Data Analyst MLOps Engineer BI Specialist
EXTERNAL SOURCES
- Wikipedia - Autoregressive integrated moving average#SARIMA
- Hyndman & Athanasopoulos - Authoritative SARIMA tutorial
- statsmodels - Python SARIMAX implementation docs
- Penn State STAT 510 - Course unit on seasonal models
TOP QUESTIONS
What is SARIMA in time series forecasting?
SARIMA is ARIMA with seasonal terms and seasonal differencing to model periodic patterns. It’s a standard statistical baseline for seasonal data.
Category: Definition
When should I use SARIMA?
Use SARIMA when data exhibit seasonality and linear dynamics. It performs well with stable seasonal patterns and moderate autocorrelation.
Category: Decision
How do I select SARIMA orders?
Use ACF/PACF plots, seasonal diagnostics, and information criteria (AIC/BIC). Grid search and automatic selection can help.
Category: HowTo
What is SARIMAX?
SARIMAX augments SARIMA with exogenous regressors, enabling the inclusion of known drivers or calendar effects.
Category: Concepts
What are common pitfalls with SARIMA?
Over-differencing can induce overdamping; ignoring residual diagnostics leads to poor forecasts. Ensure stationarity and appropriate season length.
Category: Mistakes
How does SARIMA compare to Prophet?
SARIMA is a parametric linear model, while Prophet uses additive components with robust defaults. Prophet can handle irregularities more automatically.
Category: Comparison
How do I validate a SARIMA model?
Use time-based cross-validation, residual diagnostics, and forecast accuracy metrics such as MAPE, RMSE, and MASE.
Category: Validation
Can SARIMA handle multiple seasonalities?
Classical SARIMA supports one seasonality. Multiple-season models may require TBATS or specialized approaches.
Category: Limitations
Is SARIMA suitable for real-time forecasting?
Yes, SARIMA updates quickly if parameters are fixed, but frequent re-estimation may be required for evolving series.
Category: Operations