Data Science
Informacje ogólne
Kod przedmiotu: | 2400-SZD-QPE-DS |
Kod Erasmus / ISCED: | (brak danych) / (brak danych) |
Nazwa przedmiotu: | Data Science |
Jednostka: | Wydział Nauk Ekonomicznych |
Grupy: |
Przedmioty WNE dla programu QPE w Międzydziedzinowej Szkole Doktorskiej (ZIP) |
Strona przedmiotu: | https://www.mimuw.edu.pl/~noble/courses/QPEDataScience/ |
Punkty ECTS i inne: |
(brak)
|
Język prowadzenia: | angielski |
Rodzaj przedmiotu: | nieobowiązkowe |
Skrócony opis: |
The aim of this course is to give a knowledge of and facility with modern methods of multivariate statistical data analysis: methods of density estimation, multiple regression and modern shrinkage methods for variable selection (e.g. ridge regression and LARS), cross validation techniques: dimensionality reduction (Principal Components, Canonical Correlation), discriminant analysis, modern methods for cluster analysis support vector machines. The techniques will be implemented on a wide variety of data sets using the statistical software package “R”. |
Pełny opis: |
The course deals with modern methods for multivariate statistical data analysis. The course follows, for the most part, the treatment of Alan J. Izenman. We consider 1) Nonparametric density estimation (Izenman chapter 4): optimal bin widths for histograms, kernel density estimation, projection pursuit density estimation. 2) (following Izenman chapter 5): Model assessment and selection in multiple regression. Prediction accuracy, estimating prediction error: cross validation and bootstrap techniques. Shrinkage methods for variable selection, regularised regression and least angle regression. 3) (following Izenman chapter 7) Principal component analysis, canonical variate and correlation analysis, projection pursuit. 4) Linear Discriminant Analysis (following Izenman chapter 8) binary classification, multiclass LDA, Bayes classifier, LDA via multiple regression, logistic discrimination. 5) Cluster Analysis (following Izenman chapter 12) Hierarchical methods, partitioning methods, self organising maps, clustering based on mixture models. |
Literatura: |
Alan J. IZENMAN, Modern Multivariate Statistical Techniques |
Efekty uczenia się: |
By the end of the course, the student will be familiar with a substantial toolbox of modern techniques for statistical analysis (P8S_WG) will have a broad understanding of the theoretical principles behind them (P8S_WG) will be able to implement these techniques using the statistical software R. (P8S_UW) |
Metody i kryteria oceniania: |
By project. The project should (a) be clearly written and (b) show sound basis for the choice of statistical techniques for the situation in hand. |
Właścicielem praw autorskich jest Uniwersytet Warszawski, Wydział Nauk Ekonomicznych.