Predikce časových řad

Abstract

The topic of this bachelor’s thesis is time series prediction. The objective of the thesis is to describe features of the time series, methods for processing, analysis and building various models for predicting future data, such as moving averages, exponential smoothing or ARIMA. Methods for converting to stationary time series, methods for selecting parameters for a series predictions, and methods for evaluating prediction accuracy are also described. Then the selected methods are applied to real data, and a model is built to make a forecast. Then the forecast accuracy of different models is compared.