Using Cluster Analysis Methods in Time Series Prediction

Mathematics has been developed in parallel with information technologies. This field of studies still has many objects required deep research and analysis. This paper considers the problem of time series clustering. This problem is of interest to mathematicians, financial analysts and engineers. The main problem of the classification and clustering problem is to choose appropriate function setting the distance between objects. Wrong metric can cause error and influence the result that could be crucial for the researcher. Thus, the purpose of the study is to discover the properties and effectiveness of different methods and metrics of clustering and their implementation for time series prediction. This work is organized as follows. The first chapter reviews the literature and introduces basic definitions. The second chapter includes clustering methods overview and a number of numerical experiments description. In the third chapter details an experiment on the basis of empirical data. The conclusion provides the results obtained, recommendations and referrals for the further study and research.