Course content

This course introduces chemometric methods for data analysis and experimental design, emphasising applications in biology, biotechnology, chemistry, material science, and physics. You will learn to design efficient experiments, prepare data for analysis, construct models to reveal underlying relationships in data and extract meaningful information from complex datasets.

The following topics are covered:

• Preprocessing (e.g., numerical differentiation and integration, smoothing, and frequency analysis)
• The design of experiments (e.g., full and fractional factorial design)
• Regression (e.g., linear and nonlinear least squares) and machine learning techniques for unsupervised (e.g., cluster analysis) and supervised problems (e.g., classification)
• Model validation (e.g., using test sets, cross-validation, and bootstrapping)
• Dimensionality reduction and latent variable methods to extract meaningful information from complex datasets (e.g., principal component analysis and partial least squares regression)

Learning outcome

Knowledge

After completing the course, the student can:

• Explain the principles of experimental design, including factorial designs, and how data from such designs are analyzed
• Explain the purpose and use case of various preprocessing methods
• Select appropriate chemometric techniques based on an evaluation of data analysis needs
• Describe chemometric methods for supervised and unsupervised learning (e.g., regression, principal component analysis) and provide examples of their real-world applications.
• Analyze and interpret the results obtained from chemometric methods.
• Explain how validation methods are used to evaluate the performance and accuracy of models

Skills

After completing the course, the student can:

• Reduce the dimensionality of complex datasets to extract meaningful information
• Design and analyze experiments using techniques from the design of experiments
• Prepare data for analysis using appropriate data preprocessing techniques, including cleaning, transformation, and scaling
• Perform principal component analysis and cluster analysis to explore and interpret complex datasets
• Build regression models for prediction and data analysis
• Build classification models for prediction and data categorization
• Evaluate model performance using validation techniques such as test sets and cross-validation.

General knowledge

After completing the course, the student can:

• Communicate complex chemometric results effectively, using clear and concise language and visualizations
• Apply Python programming to solve real-world chemometric problems, including data import, preprocessing, analysis, and visualization

Learning methods and activities

• Lectures.
• Exercises.
• Project work

The project work involves analysis of given data (e.g., a spectroscopic dataset from a chemical experiment), where the students make use of techniques learned in the course to solve given data analysis goals (e.g. building a predictive model for concentration of chemical compounds). This project aims to develop students' ability to apply chemometric techniques to real-world data, interpret results, and communicate findings effectively. The students must summarize their analysis in a written report, detailing their methodology, results and interpretations.

A certain number of the exercises must be approved before submitting the project report.

The expected workload for the course is 200 hours: 30 hours of lectures, 70 hours of exercises and 100 hours of independent learning, including reviewing lecture notes and project work.

Compulsory assignments

• Exercises

Recommended previous knowledge

Basic knowledge of mathematics (especially linear algebra), statistics, and chemistry (or a similar natural science field).

Course materials

The course material will be announced at the beginning of the course.

Credit reductions