EL3201 Datadriven modellering, grundläggande kurs 8,0 hp

Data-Driven Modeling, Basic Course

The course covers fundamentals of data based modeling, ranging from experiment design to estimation algorithms. The course is intended for PhD students in all areas that come in contact with data-based modeling. A common framework for most existing methods Is developed. The underlying theory as well as practical exercises are included.

  • Utbildningsnivå

  • Huvudområde

  • Betygsskala

Det finns inget planerat kurstillfälle.


After the course, the student should be able to:

·         describe the general principles for system identification.

·         identify systems in a satisfactory manner. This includes choice of excitation signals, model structure and estimation algorithm as well as proper use of model validation.

·         analyse basic model properties, such as identifiability and  accuracy (bias and variance errors).

Kursens huvudsakliga innehåll

Signal spectra, linear time-invariant sytems, prediction and filtering, linear and non-linear models, identifiability, non-parametric methods, parameter estimation, maximum likelihood estimation, linear regression, least-squares estimation, the prediction error method, the instrumental variable method, subspace identification, kernel methods, support vector machines, convergence and consistency, modeling accuracy, Cramér-Rao lower bound, numerical optimization, recursive estimation, bias and variance errors, experiment design, applications oriented system identification, choice of identification criterion, model validation, model structure selection, system identification in practice.


Lectures, exercises, presentations on selected topics by the participants, homework problems, 72 h take home exam



Lennart Ljung, System identification: Theory for the user, 2nd ed. Prentice-Hall 1999. Handouts.  

A good complement is:

Torsten Söderström and Petre Stoica. System Identification, 1989


Compulsory information


Krav för slutbetyg

·         15 min oral presentation of a selected topic in one of the lectures

·         80% on weekly home-work problems

·         project (preferably on a problem related to the student’s own research)

·         50 % on 72 h take home exam

Ges av



Håkan Hjalmarsson <hjalmars@kth.se>


Kursplan gäller från och med VT2014.