Weekly Notes for Week 28

This week is about principal component analysis (PCA), principal component regression (PCR), molecular descriptors, and QSAR.

• Topics:

  • PCA as a change of basis for data matrices.
  • Centering, covariance matrices, eigenvectors, scores, and loadings.
  • Dimensionality reduction and visualisation.
  • From PCA to PCR: regression in principal-component score space.
  • Molecular descriptors and QSAR as cheminformatics applications.

• Slides and reading:

  • PCA and Eigenfaces
  • Molecular descriptors
  • PCA reading notes

• Exercises and code:

  • Exercise Sheet 08: PCA and PCR
  • PCA projection visualisation script