The publication of the lecture course "Time Series Analysis"continues in this issue. In the previous issue, the basic definitions of the concept of stationary random processes, their characteristics and properties, as well as the class of stationary random processes such as ARMA and non - stationary-type ARIMA were considered. Analyzed the approach of Box-Jenkins to the identification of the time series.  This issue continues the discussion of the approach of Box-Jenkins, in particular, the estimation of the parameters of the type of ARIMA and forecasting using these models, examples of applying the approach of Box-Jenkins.  The features of the behavior of some non-stationary time series, non-stationary series such as TS and DS, Dickey - fuller test to test the hypothesis about the type of series are considered.