Course - Multivariate analysis and Machine learning methods - TTK4260
TTK4260 - Multivariate analysis and Machine learning methods
Theory and applications of data analytical methods that are instrumental for the work of control engineers (see the intended learning outcomes for a summary of which ones). Overview and demonstration of the capabilities of other machine-learning oriented data analysis methods (again, listed in the intended learning outcomes). Discussions about how to structure the data analysis workflows, how to interpret the data and the results, and how to contextualize the various approaches so to be able to select the appropriate ones and motivate the selection. The course will thus have two distinct "working modes": one where the theory of the algorithms will be presented in detail, and one where the algorithms will be introduced and demonstrated without deriving them in detail. The first part deals with tools that are within the core knowledge that control engineers shall have. The second deals with ancillary tools and provides an overview of the possibilities offered by the current state-of-the-art methods within Machine Learning.
Intended cognitive learning outcomes
* Refreshing background knowledge:
- motivations and underlying points of view
- overview of the data analysis methods types and traditions
- Least Squares
- Maximum Likelihood and Maximum a posteriori
- statistical performance indexes
- bias vs variance trade-off
* Detailed analysis of the following algorithms / methods:
- PCA, ICA, PCR, PLS, multiblock and PARAFAC algorithms
- outlier detection
- time series prediction
- time series models identification
- model order selection
- design of experiments
* Overview and demonstration of the capabilities / working strategies of:
- change detection
- IDLE methods
- subspace identification
- Neural networks
- Random Forests
- Support vector machines
- other clustering methods (especially Nearest Neighbours)
- other classification methods
Intended non-cognitive learning outcomes:
- understand the philosophies, strengths, and limitations of the various methods
- knowing the meaning of the data and the interpretation of the data
- contextualising the learned strategies and understanding how to combine them
- become independent, self-confident, and critical with regards to data analysis
 K. Poynton, Cognitive and non-cognitive learning factors, http://cim.acs-schools.com/wp-content/uploads/2015/08/Cognitive-and-non-cognitive-learning-factors.pdf
Learning methods and activities
- both frontal lectures and flipped classrooms
- peer instruction sessions
- interactive data analysis sessions
- at-home data analysis projects on some preassigned datasets
- compulsory exercise
Further on evaluation
The written exam provides the basis for the final grade in the course. In case of postponed examination (continuation examination), the written examination may be changed to oral examination. If, after the postponed examination, the student has not yet passed the exam, the student must repeat the entire course the next academic year.
Compulsory activities from previous semester may be approved by the department.
Recommended previous knowledge
Basic knowledge about probability theory, linear systems, and ordinary differential equations.
Required previous knowledge
Good understanding of basic linear algebra.
The course material will be presented at the start of the course
Credits: 7.5 SP
Study level: Second degree level
Term no.: 1
Teaching semester: SPRING 2022
Language of instruction: English
- Signal Processing
- Design of Experiments
- Engineering Cybernetics
- * The location (room) for a written examination is published 3 days before examination date. If more than one room is listed, you will find your room at Studentweb.
For more information regarding registration for examination and examination procedures, see "Innsida - Exams"