Computational Statistics
Títol del curs: Computational StatisticsImpartit per: Dimitris Karlis
Llengua del curs: Anglès
Dates i horaris del curs: 4-8 juliol 2011, de 10 a 12 h.
Lloc: FME, dilluns, dimarts i dijous, aula 100, dimecres i divendres aula PC1
Tipus d'activitat i càrrega lectiva: Curs de 10 hores
Reconeixement acadèmic: 1,5 crèdits
Data de matrícula: 6-26 de juny de 2011
Course Objective:
The objective of this course is to teach the student how to apply computational methods for statistical inference and in particular Bootstrap and Monte Carlo methods to estimate standard errors, construct confidence intervals, test hypotheses in a variety of statistical problems.
Course Content:
The course will cover computer intensive methods for statistical inference, including Monte Carlo methods, Bootstrap and other resampling methods. There will be a mixture of theoretical justification of the methods and applications to a wide variety of problems in statistics, discussing the advantages and the disadvantages of the approaches and giving a comparative treatment of the methods. The topics that will be covered are:
- The idea of Bootstrap
- Parametric versus non-parametric bootstrap.
- Bootstrap estimates of standard errors and bias.
- Bootstrap confidence intervals.
- Variants of bootstrap.
- Bootstrap for complicated structures.
- Other resampling schemes including jackknife.
- Real data applications with different kinds of data.
- Monte Carlo tests.
There will be two computer sessions where the theory will be applied using R package and real data examples.
Evaluation system:
There will be a project that the students must submit.
Selected bibliography:
Chernick, M. R. (1999) Bootstrap Methods: A practitioner’s Guide.
Willey Davison, A.C. and Hinkley, D.V. (1997) Bootstrap Methods and Their
Applications. Cambridge University Press, Cambridge.
Documents:
Share: