Autoregressive Integrated Moving Average (ARIMA )

Introduction: ARIMA

ARIMA is a statistical model used for analysing and forecasting time series data. The acronym stands for Autoregressive Integrated Moving Average. It is commonly applied in economics, finance, and operations where data show patterns over time.

Background

The ARIMA model was popularised by statisticians George Box and Gwilym Jenkins in the 1970s. It remains a cornerstone of time series forecasting and is widely known as the Box-Jenkins methodology.

Key Elements / Features

• Autoregressive (AR): Explains the variable using its own previous values.
• Integrated (I): Uses differencing to make the data stationary, meaning mean and variance stay stable over time.
• Moving Average (MA): Models forecast errors as a combination of past error terms.

Parameters

An ARIMA model is defined by three parameters:

• p: Number of autoregressive terms.
• d: Number of non-seasonal differences required to achieve stationarity.
• q: Number of lagged forecast errors in the moving average component.

Applications / Examples

• Finance: Forecasting stock prices or exchange rates.
• Economics: Predicting GDP, inflation, or unemployment rates.
• Operations management: Demand forecasting and inventory planning.

For example, a retailer can apply ARIMA to sales data to forecast demand for the coming months, ensuring stock levels match customer needs.

Relevance / Impact

ARIMA is valued for its flexibility and accuracy with stationary data. However, it requires careful preprocessing and parameter tuning. For strongly seasonal patterns, the SARIMA (Seasonal ARIMA) variant is often more suitable. Despite its limitations, ARIMA remains one of the most widely used forecasting models in research and business.

See also

• SARIMA (Seasonal ARIMA)
• Time Series Analysis
• Forecasting Methods
• Box-Jenkins Methodology