Quantitative data analysis [ LLSMF2013 ]
5.0 crédits ECTS
30.0 h
2q
| Teacher(s) |
Saerens Marco ;
|
| Language |
English
|
Place
of the course |
Louvain-la-Neuve
|
| Main themes |
Presentation of quantitative data analysis methods, in particular scoring methodology and classification;
Presentation of some decision making models;
Reading texts containing data analysis methods;
Exercises in appropriation by a group work, in analysing methods of qualitative and quantitative materials collected personally or placed at the disposal;
Initiation to professional data analysis software such as Atlas-TI, SAS/JMP and R.
|
| Aims |
To be able to make decisions on the basis of quantitative information, and to assess accurately the performances of the mobilized models.
|
| Content |
Content
The study of data analysis and decision-making methods, with a focus on the interpretation of the results; in particular, classification, scoring methodology: clustering, factorial and projection methods, decision trees, logistic regression,
A discussion on which method to use in function of the problem at hand and the available data.
Methods
A combination of lectures, practical exercises and a project dealing with real data.
Content
A review of the main subspace projection and feature extraction of data analysis/modeling, and their interpretation:
- Categorical data: subspace projection and latent variable techniques techniques, log-linear models, etc.
- Numerical data: subspace projection and latent variable techniques, clustering techniques, discriminant analysis, etc.
Supervised classification: naïve Bayes, artificial neural networks, decision trees, combining classifiers, etc.
Unsupervised classification (clustering) methods.
Decision-making from data: a short introduction to Bayes decision theory, Bayesian networks, Markov decision processes, reinforcement learning, multicriteria decision analysis.
Application to "information retrieval" and to "web mining" (PageRank, Hits, collaborative recommendation, etc).
A discussion of which method to use in function of the data and the problem at hand.
Projects (for instance scoring) based on real data, with SAS/JMP, S-Plus or R.
Methods
In-class activities
0 Lectures
0 Project based learning
At home activities
0 Readings to prepare the lecture
0 Paper work
|
| Other information |
Prerequisites (ideally in terms of competencies): A course in multivariate statistical analysis, on probability theory, on mathematical statistics, on matrix algebra and on multivariate analysis.
Evaluation :
Writing of two papers.
Support : Book chapters provided to the students
References : Provided during the class
- Duda, Hart & Stork (2001), "Pattern classification, 2nd ed". John Wiley & Sons.
- Bardos (2001), "Analyse discriminante. Application au risque et scoring financier. Dunod.
- Lebart, Morineau & Piron (1995), "Statistique exploratoire multidimensionnelle". Dunod.
- Webb (2002), "Statistical pattern recognition, 2nd ed". John Wiley and Sons.
- Theodoridis & Koutroumbas (2003), "Pattern recognition". Academic Press.
- Alpaydin (2004), "Introduction to machine learning". MIT Press.
- Han & Kamber (2000), "Data mining: Concepts and techniques". Morgan Kaufmann.
- etc.
|
Cycle et année
d'étude |
> Master [60] in History
> Master [120] in History
> Master of arts in Business engineering
|
Faculty or entity
in charge |
> CLSM
|
<<< Page précédente
|
