Abstract:
An algorithm for forecasting non-stationary time series based on the analog method is considered. The proposed method is tested at air temperature values according to an automatic meteorological station. Six forecast models using various measures of similarity were constructed, the forecast error was determined on an independent sample. Comparison with the estimate of the inertial prediction led to the conclusion that the method of analogs gives satisfactory results of the current temperature prediction.
Keywords:analogue method, current temperature forecast, proximity measures.