A characteristic of time-series models (the I in ARIMA models) in which one or more of the differences of the time-series data are included in the model. The term integrated is used because the original series may be recreated from a differenced series by summation.