What is ARIMA?

An Autoregressive Integrated Moving Average (ARIMA) model is a widely used time series forecasting technique. Autoregressive models use a linear combination of data from previous time steps to predict future values, while Moving Average models use a linear combination of past forecast errors. ARIMA models combine both of these approaches. They’re also Integrated, which means that they use the differences between data points rather than the data points themselves, in order to remove trends such as moving averages changing over time.

The statistics behind ARIMA models call for a time series to be regular and stationary, which is what requires the elimination of trends. You can use seasonal differencing to remove seasonal components from time series, or use more complicated models that include seasonality. Confidence intervals for ARIMA models assume that residuals are normally distributed and uncorrelated, and it’s important to check this when using these models. ARIMA models are designed for metric data collected at regular time steps and can be helpful tools for modeling financial data. There are tools and packages for creating ARIMA models in many languages including Python, R, and JavaScript.

Related resources