Analysis and Forecasting of Co-integrated Time-series

The thesis is dedicated to the topic of analysis and forecasting of cointegrated time series. In the first chapter a theoretical background of the research is given, including main definitions and literature analysis, which introduces a modern methodological framework, referring the questions of time series forecasting and usage of cointegration concept in different fields. In the second chapter main time series forecasting models, as well as methods of forecasting cointegrated time series, along with well-known performance metrics, are described. In a separate chapter an overview of existing data analysis tools is provided, the results of an experiment aimed at forecasting cointegrated time series on a test dataset of Air Quality Indices (AQIs) for five air pollutants for one of the major cities of Asia-Pacific region for two years are given. The paper can be of interest for researchers and organizations concerned with time series (and in particular cointegrated time series) forecasting, as well as for life science companies.