ARIMA

← Back to the article

ARIMA looks at past numbers in order and uses them to guess future numbers.

What it does

ARIMA looks at past numbers in order and uses them to guess future numbers.

Time Series: Data in time order • Seasonality: Repeating cycles

How it works

  • Read the history of values.
  • Look for patterns and changes.
  • Forecast the next values.

Uses and applications

  • A shop can estimate next week’s sales.
  • A power company can predict electricity use.
  • A weather station can forecast a simple time pattern.

In short

ARIMA forecasts using time-ordered data.

Related methods

  • Prophet: Adds seasonality and holiday effects to forecasting.
  • Linear Regression: Fits a line to features rather than to past values in order.
  • LSTM (Long Memory): A neural network that can also learn from sequences.

Back to Forecasting and optimization · All topics

The interactive demonstration and the Python walkthrough for this topic need JavaScript.