Stampa la pagina Condividi su Google Condividi su Twitter Condividi su Facebook Statistical Modeling

Professor Alfonso Iodice D'Enza

ECTS: 5

Prerequisites: None

Contents:

Simple linear regression: estimating coefficients; assessing the accuracy of the estimates and of the model. Multiple linear regression. Qualitative predictors. Extensions of the linear model. Potential problems. Linear regression and K-nearest neighbors regression. Non-linear models for regression. Polynomial regression; step functions; basis functions, regression splines. Smoothing splines. Local regression. Generalized additive models.

Textbooks:

An Introduction to Statistical Learning, with application in R. G. James, D. Witten, T. Hastie and R. Tibshirani. Freely downloadable here

http://www-bcf.usc.edu/~gareth/ISL/getbook.html

Aims:

The course aim is to provide a modern approach to statistical models, with a special focus on regression problems. Linear and non-linear models are defined in light of the bias-variance trade-off and of the flexibility-interpretability trade-off. All the methods are introduced from both a theoretical and applicative perspective, and compared to one another. Model selection and model performance assessment will be also addressed. All the covered topics will be implemented in Cran-R meta-language.

Teaching:

Lectures and lab sessions.

Examination methods:

Final project dissertation.

n.

[Ultima modifica: mercoledì 30 novembre 2016]