Multivariate Data Analysis - Advanced Topics

• Design of Experiments
• Principal Component Analysis
• Multivariate regression methods (MLR,PCR,PLSR)
• Strategies for model selection and validation (bias-variance trade-off)
• Features and variables selection
• Classification methods (Machine learning)
• Tree-based methods
• ANN/CNN
• Time series analysis
• Prediction Error Methods for the Identification of dynamical systems
• Kalman filters
• Metamodelling & hybrid modelling
• Independent Component Analysis
• PARAFAC, multiblock (sensor fusion) and IDLE modelling

KNOWLEDGE: The students shall get an overview of different methods for analysing data from processes that are continuous and/or time dependent, both for exploratory data analysis, quantitative prediction and classification. They shall be able to plan practical experiments using statistical principles. This includes sensor-fusion and hierarchical modelling of multiblock data. SKILLS: The students shall be able to organize data form different types of measuring instruments, with different dimensions and consider optimal pretreatment of data. They shall be able to propose the most suitable methods given a specific application. GENERAL COMPETENCE: Be able to use knowledge and skills on new applications. Be able to discuss topics related to the course with specialists in the topics and propose which methods from the course to use in interdisciplinary projects.

Lectures, Group work on various topics. Project work on chosen datasets.