. – Mathematical statistics II PROFESSOR LUCIO BERTOLI BARSOTTI Text under revision. Not yet approved by academic staff. COURSE AIMS To introduce the main paradigms of statistical inference, with particular attention to point and interval estimation and hypothesis tests. COURSE CONTENT - Multivariate random variables. Bivariate random variables, discrete and continuous. Marginal. Conditional. Correlation and independence. Conditional expected value. Regression function. Multinomial random variable. Bivariate normal random variable. Multivariate normal. - Transformations of multivariate random variables. General case. Sum. Linear combinations of the components of a multivariate normal. - Parametric families: location-scale family and exponential family. - Sampling and random variables. Sample space. Determination of the distribution function of a random variable when using simple random sampling. Likelihood function. - Statistical information. Overview. Sample average and variance. Exact and approximate distributions of sample moments. Fisher information. Cramér–Rao information inequality. Sufficiency and ancillarity. Neyman Fisher factorization criterion. Subordinate and equivalent statistics. Minimal sufficiency. Completeness. Sufficient statistics and the exponential family. Rao-Blackwell theorem. - Parametric estimation. Identifiability. Estimator and estimation. Comparison of estimators based on the mean square error. Unbiasedness, consistency, efficiency and asymptotic efficiency. - Estimation methods. Method of moments. Maximum likelihood method (ML). ML estimation in the exponential family case. Asymptotic optimality of ML estimators. - Confidence intervals (CIs). Construction in the general case. Pivotal quantity method. Calculation of minimum sample size needed for a certain CI size. Applications to cases of determination of exact CIs in normal sampling. Asymptotic confidence intervals: applications to cases of continuous and discrete random variables. Determination of exact CIs in the case of discrete random variables: general procedure. Confidence intervals for comparing two normal populations on the basis of two simple random samples of different sizes: confidence interval for the difference in means and confidence interval for the comparison of variances. - Statistical hypothesis testing theory. Exact parametric tests for the mean and variance in normal distribution. Asymptotic tests for parameters in non-normal models. Nonparametric tests: Kolmogorov-Smirnov test. Chi-squared test. Rank tests. READING LIST L. BERTOLI - BARSOTTI, Corso di Statistica Matematica, Quaderni del Dipartimento di Matematica, Statistica, Informatica e Applicazioni Università di Bergamo, Serie Didattica, n. 3, 2005. A.M. MOOD - F.A. GRAYBILL - D.C. BOES, Introduzione alla Statistica, Mc-Graw-Hill, 1991. N. WEISS, Calcolo delle Probabilità, Pearson PBM, 2008. L. PACE, A. SALVAN, Introduzione alla Statistica Vol. 2 – Inferenza, Verosimiglianza, Modelli, Cedam, 2001. PELOSI, SANDIFER, CERCHIELLO, GIUDICI, Introduzione alla Statistica 2nd edition, Mc-Graw-Hill, 2009. TEACHING METHOD Theory lessons plus a cycle of exercises (first unit only). ASSESSMENT METHOD Preliminary written examination followed by an oral examination. NOTES Further information can be found on the lecturer's webpage http://www2.unicatt.it/unicattolica/docenti/index.html or on the Faculty notice board. at