FSI-SSPCompulsoryMaster's (2nd cycle)Acad. year: 2016/2017Summer semester1. year of study4 credits
The course provides the introduction to the theory of stochastic processes. The following topics are dealt with: types and basic characteristics, covariation function, spectral density, stationarity, examples of typical processes, time series and evaluating, parametric and nonparametric methods, identification of periodical components, ARMA processes. Applications of methods for elaboration of project time series evaluation and prediction supported by the computational system MATLAB.
Learning outcomes of the course unit
The course provides students with basic knowledge of modelling of stochastic processes (decomposition, ARMA) and ways of estimate calculation of their assorted characteristics in order to describe the mechanism of the process behaviour on the basis of its time series observations. Students learn basic methods used for real data evaluation.
Mode of delivery
90 % face-to-face, 10 % distance learning
Rudiments of the differential and integral calculus, probability theory and mathematical statistics.
Recommended optional programme components
Recommended or required reading
Cipra, Tomáš. Analýza časových řad s aplikacemi v ekonomii. 1. vyd. Praha : SNTL - Nakladatelství technické literatury, 1986. 246 s.
Brockwell, P.J. - Davis, R.A. Time series: Theory and Methods. 2-nd edition 1991. Hardcover : Corr. 6th printing, 1998. Springer Series in Statistics. ISBN 0-387-97429-6.
Hamilton, J.D. Time series analysis. Princeton University Press, 1994. xiv, 799 s. ISBN 0-691-04289-6.
Anděl, Jiří. Statistická analýza časových řad. Praha : SNTL, 1976.
Ljung, L. System Identification-Theory For the User. 2nd ed. PTR Prentice Hall : Upper Saddle River, 1999.
Brockwell, P.J. - Davis, R.A. Introduction to time series and forecasting. 2nd ed. New York : Springer, 2002. xiv, 434 s. ISBN 0-387-95351-5.
Planned learning activities and teaching methods
The course is taught through lectures explaining the basic principles and theory of the discipline. Exercises are focused on practical topics presented in lectures.
Assesment methods and criteria linked to learning outcomes
Graded course-unit credit requirements: active participation in seminars, demonstration of basic skills in practical data analysis on PC, evaluation is based on the oral exam and outcome of an individual data analysis project.
Language of instruction
The course objective is to make students familiar with principles of theory stochastic processes and models used for analysis of time series as well as with estimation algorithms of their parameters. At seminars students practically apply theoretical procedures on simulated or real data using the software MATLAB. Result is a project of analysis and prediction of real time series.
Specification of controlled education, way of implementation and compensation for absences
Attendance at seminars is controlled and the teacher decides on the compensation for absences.
Type of course unit
26 hours, optionally
Teacher / Lecturer
1. Stochastic process, types, trajectory, examples.
2. Consistent system of distribution functions, strict and weak stacionarity.
3. Moment characteristics: mean and autocorrelation function.
4. Spectral density function (properties).
5. Decomposition model (additive, multiplicative), variance stabilization.
6. Identification of periodic components: periodogram, periodicity tests.
7. Methods of periodic components separation.
8. Methods of trend estimation: polynomial regression, linear filters, splines.
9. Tests of randomness.
10.Best linear prediction, Yule-Walker system of equations, prediction error.
11.Partial autocorrelation function, Durbin-Levinson and Innovations algorithm.
12.Linear systems and convolution, causality, stability, response.
13.ARMA processes and their special cases (AR and MA process).
seminars in computer labs
13 hours, compulsory
Teacher / Lecturer
1. Counting, storage and visualization of data, moment characteristics of stochastic process.
2. Simulating time series with some typical autocorrelation functions: white noise, coloured noise with correlations at lag one, exhibiting linear trend and/or periodicities.
3. Detecting heteroscedasticity. Transformations stabilizing variance (power and Box-Cox transform).
4. Identification of periodic components, periodogram, and testing.
5. Use of linear regression model on time series decomposition.
6. Estimation of polynomial degree for trend and separation of periodic components.
7. Denoising by means of linear filtration (moving average): design of optimal weights preserving polynomials up to a given degree, Spencer's 15-point moving average.
8. Filtering by means of stepwise polynomial regression.
9. Filtering by means of exponential smoothing.
11.Simulation, identification, parameters estimate and verification for ARMA model.
12.Testing significance of (partial) correlations.
13.Tutorials on student projects.