300053 SE Principles of multivariate statistics for biologists (2016S)

The course should represent an introduction to the theoretical principles of selected multivariate-statistical methods for the analyses of biometrical data, and is mainly intended for advanced students with previous knowledge in biometry and basic statistics. It is planned that the representations of the theoretical principles will be followed by the practical work of autonomously creating computer programm macros under technical guidance ('learning by doing'; preferred primary computer packages: Openoffice calc, Gnumeric). Out from the following list some topics will be explained in detail: biometric variables, data transformations, variance, covariance and correlation, multivariate representation of variance-covariance and correlation (matrices, vectors), simple and multiple linear regression analysis, partial regression and correlation, linear compounds, character space - Euclidean vs. MAHALANOBIS distances, principal components analysis (e.g., eigenvector, eigenvalue, total variance), factor analyses, factor space, ordination techniques, linear discriminant analyses, discriminant space, identification analysis.