A time series is said to be stationary if its statistical characteristics do not change over time. That is, if it exhibits the same mean, variance and autocorrelation structure over time; otherwise, the time series is said to be nonstationary.