Statistical Learning

Jean-Philippe Vert

Université Gaston Berger, Saint-Louis, Senegal
February 14-23, 2011

This course introduces concepts and methods for statistical machine learning, in particular regression and pattern recognition. It is sponsored by the STAFAV project.

Results and solution of the exam (exercise 4.2 of ELS2)

Notes

We will mainly follow a few chapters of the book The elements of statistical learning, by T. Hastie, R. Tibshirani and J. Friedman. The book is freely available on the book website

Slides (summary)

Schedule

DateLectureMaterial
Monday 14/2Introduction to statistical learning, linear regression, k-NN
Tuesday 15/2Model selection, cross-validation: practicalPractical: k-NN and cross-validation
Thursday 17/2Linear regression: least squares, feature subset selection
Friday 18/2Linear regression: ridge regression, Lasso, PCR, PLS
Monday 21/2Linear regression: practicalPractical: linear regression
Tuesday 22/2Linear classification: LDA, logistic regression (LR), regularized LR
Wednesday 23/2Linear classification: practicalPractical: linear classification

Practical sessions

The practical sessions require R, with packages leaps, MASS, lars, pls, glmpath, e1071.
We will use the data prostate cancer and South African heart disease.

Validation

  1. Faire exercice 4.2 du livre (lien entre LDA et OLS)
  2. Choisir un jeu de données et tester différentes méthodes de régression ou classification binaire vues dans le cours. Diviser les données en train/test, et comparer les différentes méthodes entraînées sur le train par leur performance sur le test. Rendre le code en R et un rapport de 2-3 pages.
Envoyer le tout à Jean-Philippe.Vert@mines.org avant le 1er avril 2011.

Back to my homepage