Time Series Analysis
Note: The time of the lecture has rescheduled to Thursday 8:30 - 10:00, venue of the first lecture (15.10.2015) is Heho18, E 60, as of the second lecture (22.10.2015) Heho18, E20.
2h lectures + 2h exercises
Time and Venue:
Lecture: Thursday 8:30 - 10:00. Room: 15.10.2015 Heho 18, E60, as of 22.10.2015 Heho 18, E20.
Exercises: Thursday 14:15 - 15:45 in Heho 18 room E20, first exercise class: 22.10.2015.
Probability Theory, Statistics
In many application areas, the data to be analyzed form a sequence of observations given at a sequence of time points, that is, a time series. For instance, stock prices, exchange rates or meteorological data are typically recorded at a sequence of time points and thus yield time series. The fact that the data are subject to a certain chronological order is crucial for their analysis and has to be taken into account when formulating statistical models. Trends, seasonal effects, and stationarity will be fundamental notions in this course. We will discuss autocovariance and autocorrelation functions as a tool for analyzing dependencies in time. Particular attention will be given to ARMA (auto regressive moving average) processes as the most important linear model for time series. Within the setting of ARMA processes we will discuss statistical inference and forecasting methods. In addition to problems on the mathematical theory, homework sets will include practical examples. The course will be taught in English.