SARIMA

Introduction: SARIMA (Seasonal ARIMA)

SARIMA, or Seasonal Autoregressive Integrated Moving Average, is an advanced statistical model used for analysing and forecasting time series data that exhibit regular seasonal patterns. It extends the traditional ARIMA model by incorporating seasonal components, making it particularly effective for data with repeating cycles, such as monthly sales, quarterly production, or annual temperature trends.

Background

While the ARIMA model performs well with stationary (non-seasonal) time series, many real-world datasets include strong seasonal effects. For example, retail sales often rise during holidays, and electricity usage peaks in certain months. To address this, the SARIMA model—developed as part of the Box-Jenkins methodology—adds seasonal terms that capture these recurring patterns. This enables more accurate forecasts for processes influenced by cyclical trends and predictable fluctuations.

Key Elements / Features

SARIMA is typically written as:

• SARIMA (p, d, q) (P, D, Q) m

where:

• p, d, q = Non-seasonal ARIMA parameters (autoregressive, differencing, moving average)
• P, D, Q = Seasonal counterparts for autoregression, differencing, and moving average
• m = Number of periods in each season (e.g., 12 for monthly data with yearly seasonality)

Key features include:

• Seasonal Autoregressive (P): Captures relationships between observations separated by full seasonal periods.
• Seasonal Differencing (D): Removes repeating seasonal trends to stabilise the data.
• Seasonal Moving Average (Q): Models seasonal forecast errors over time.

Applications / Examples

• Retail: Forecasting monthly sales to prepare for holiday peaks.
• Tourism: Predicting hotel occupancy across high and low seasons.
• Energy: Modelling electricity or gas demand influenced by temperature changes.

Example: An airline company uses SARIMA to forecast passenger demand, accounting for recurring spikes during summer and winter holidays.

Relevance / Impact

SARIMA provides a powerful framework for forecasting time series data influenced by both trend and seasonality. It improves planning accuracy in industries where demand fluctuates cyclically, helping organisations optimise resources, inventory, and capacity. By combining ARIMA’s statistical rigor with seasonal adjustments, SARIMA delivers a balanced and reliable forecasting method for real-world applications.

See also

• ARIMA (Autoregressive Integrated Moving Average)
• Time Series Analysis
• Lean Six Sigma
• Box-Jenkins Methodology