Course content

The course is an introduction to applied chemometrics and very relevant to data analysis within chemoinformatics, bioinformatics and process chemistry.
Methods covered are: Classical least squares method and polynomial regression. Experimental design method such as full and fractional factorial design. Preprocessing methods such as normalisation, Fourier filtering, Savitsky-Golay numerical differentiation, convolution, multiplicative scatter correction, autoscaling. Latent variable methods such as principal component analysis/singular value decomposition, principal component regression, partial least squares regression. Model validation using test set, cross validation, bootstrap and y-randomisation. Cluster analysis techniques such as hierarchical cluster analysis, k-means cluster analysis and «gap» statistic. Classification methods such as random forest, CART and k-nearest neighbour method. Multivariate curve resolution methods such as evolving factor analysis, alternating least squares and local rank analysis.

Learning outcome

After completing the course, the student should be able to:
(1) Set up, analyse and interpret results from an experimental design
(2) Choose what type of preprocessing methods are necessary in different situations
(3) Make use of unsupervised methods such as PCA and cluster analysis to analyse and interpret multivariate data sets
(4) Make use of regression methods such as PLS and PCR to analyse, interpret and predict multivariate data sets
(5) Make use of validation methods to check the predictive ability of various models
(6) Be able to explain the strengths and weaknesses of both supervised and unsupervised methods

Learning methods and activities

Lectures. Computer exercises and calculating exercises.

Recommended previous knowledge

Basic level in mathematics (in particular linear algebra) and chemistry.

Course materials

Is given in the beginning of the course.

Credit reductions