method for constructing the stochastic model based on regression and spectral analysis

The article gives a procedure for constructing a stochastic model, whose efficiency was verified on hydrological data. The procedure includes the construction of regression relationships between the examined variable and its increment, the removal of hidden periodicities, the reduction of the residual sequence to a normal distribution with the use of Johnson transformation, and testing a hypothesis regarding the independence of the residual sequence. The reduced model was used to obtain a probabilistic forecast of anticipated events.