FDMS: "Fouille de Données et Médias Sociaux" course at Université Pierre et Marie Curie.
Practical sessions:
- Data science tutorial: predict Titanic survivors, see the article on my website.
- Playing with fishes: polynom interpolation, cross-validation, Akaike information criterion on fish data
- L1-regularization and parcimonious models
- Bagging and boosting implementations
- Data visualization and recommendation: t-SNE and collaborative filtering
- SoRec: collaborative filtering for movie recommendation
- Collective inference: classification in graphs, labelling unknown university pages
- Diffusion: How do diseases/trends propagate?
- Bandit learning, linear UCB: ad targetting