Course for international guest/part time students
- Faculty
- Faculty of Science
- Organization
- TTK Department of Probability Theory and Statistics
- Code
- tdimst1u0um17em
- Title
- Multivariate statistical methods (l)
- Usual semester
- Spring
- Published semester
- 2025/26/2
- ECTS
- 6
- Language
- hu
- Learning outcomes
- Knowledge: getting familiar with the modern notions of multivariate statistical methods Ability: to understand and use multivariate statistical methods Attitude: the need to extend the mathematical knowledge, to gain new analytic and applied programming skills Autonomy and Responsibility: based on the gained knowledge in multivariate statistical methods, the students are able to decide which tools are the most suitable to solve applied problems
- Course content
- Estimation of the parameters of the multidimensional normal distribution. Matrix-valued distributions. The Wishart distribution: its density function, determinant, expected value of its inverse. Hypothesis testing of the parameters of the multidimensional normal distribution. Independence testing. Normality testing. Linear regression. Measurement of the relationship between variables: correlation coefficient, maximum correlation, partial correlation, canonical correlation. Principal component analysis, factor analysis, variance analysis. Discrete, multivariate models, Contingency tables. Maximum-likelihood estimation in a loglinear model. Kullback-Leibler divergence. Linear and exponential distribution families. Numerical determination of the L-projection (Csiszár's method, Darroch-Ratcliff procedure).
- Assessment method
- exam
- Bibliography
- lecture notes