Description: In many applications data is acquired from multiple sources or is described by multiple variables. Within areas of signal and image processing, biomechanics, clinical decision support, bioinformatics, etc. there is an increasing need for analysing various types of multivariate data containing both dependent (outcome) and independent (predictor) variables. Instead of analysing the data by treating it as a task of multiple univariate problems the methods in this course fully encompasses the task as a single multivariate problem. In effect, the covariate nature of the data is analysed by examining relationships among multiple variables at the same time. The content of the course will cover both representation, transformation, and visualization of multivariate data as well as multivariate methods for statistical analysis, pattern recognition and test evaluation. The tentative course program is:

• Probabilistic classification
o Bayesian decision theory
o Multivariate density
• Non-parametric classification
o Linear discriminant functions
o Support Vector Machines
o Clustering
• Data analysis and test evaluation
o Data representation, visualization, transformation (PCA/ICA)
o ROC analysis
o Cross validation
• Multivariate statistics
o Multivariate analysis of variance
o Multivariate regression analysis

Literature
Duda, Hart, Stork, Pattern Classification, 2nd Edition, Wiley
Jerrold H. Zar, Biostatistical Analysis, 5th Edition, Pearson (or alternative literature)

Prerequisites
Statistics, Matlab (basic skills), SPSS (basic skills).

Evaluation
Evaluation of the course will be based on active participation in the course including practical exercises in Matlab and SPSS.

Organizer: Associate Professor Lasse Riis Østergaard, e-mail: lasse@hst.aau.dk and Associate Professor Carsten Dahl Mørch, e‐mail: cdahl@hst.aau.dk

Lecturers: Associate Professor Lasse Riis Østergaard and Associate Professor Carsten Dahl Mørch, Aalborg University

ECTS: 2.0 (subject to changes)

Time: 26 October, 2, 10 and 23 November 2015 (8.15-12.00)

Place: Aalborg University, Niels Jernes Vej 14, 3-119

Zip code: 9220

City: Aalborg

Number of seats:

Deadline: 5 October, 2015