Lecturer
Prof. Dr. Sc. LAVOSLAV ČAKLOVIĆ
Department of Mathematics, Faculty of Science, University of Zagreb
Date and time
Wednesday, February 15, 2012 at 3:30 p.m.
Location
Lecture Hall, Wing I, Ruđer Bošković Institute, Bijenička 54, Zagreb
Abstract
Multivariate methods used in statistical analysis will be presented, which are becoming more and more interesting in the context of 'data mining', modeling dependencies between variables (especially categorical) and extracting informational features necessary for object recognition, among which:
‐ Principal Component Analysis (PCA)
‐ Correspondence Analysis (simple (CA) and canonical (CCA))
- Valuation methods (with an emphasis on the potential method)
The common characteristic of the above methods is the ordering of a set of objects on a selected scale. Depending on the type and quality of the input information, one is decided upon.
(or more) than data analysis methods. The scale additionally serves to calculate the measure of similarity between objects and their categorization/clustering. The methods are illustrated using examples from everyday practice with an emphasis on diagnostics in medicine.
All calculations and analyzes were carried out with the help of the statistical package R and Mathematica.
