Syllabus

18.655 - Mathematical Statistics

We will cover the following topics:

  • Statistical decision theory

  • Exponential families

  • Data reduction: sufficiency, ancillarity, completeness

  • Bayes and minimax estimation

  • Hypothesis testing and confidence intervals

  • Maximum likelihood estimation

  • Large sample theory

  • Resampling methods: bootstrap and permutation tests

  • Sequential testing

A tentative schedule is as follows. Reading assignments are from Keener, unless otherwise specified. W - High Dimensional Statistics by Wainwright (see references)

Date Topic Reading
2/6 Introduction, probability models, risk functions 1, 3.1
2/11 Exponential families 2
2/13 Sufficient statistics 2, 3.2
2/19 (Monday schedule) Minimal sufficiency, completeness 3.4, 3.5, 3.6
2/20 Rao-Blackwell theorem, UMVU estimation 3.6, 4.1, 4.2
2/25 UMVU estimation, Cramer-Rao 4.2, 4.2, 4.5
2/27 Information and estimation 4.5, 4.6
3/4 Bayesian estimation 7.1, 7.2
3/6 No class -
3/11 Hypothesis testing, Neyman-Pearson lemma 12.1, 12.2, 12.3, 12.4
3/13 UMP tests 12.3, 12.4, 12.5, 12.7
3/18 UMP unbiased tests 13.1, 13.2, 13.3
3/20 Midterm (in class) -
3/25 Spring break -
3/27 Spring break -
4/1 Asymptotic concepts 8.1, 8.2, 8.3,
4/3 Maximum likelihood 8.3, 8.4
4/8 Relative efficiency 8.5, 9.1, 9.2 Stigler, Efron
4/10 Consistency of MLE 9.1, 9.2, 9.3
4/15 Holiday - Patriot's day -
4/17 Asymptotic normality of MLE 9.1, 9.2, 9.3
4/22 Asymptotic tests and CIs 9.4, 9.5, 9.7
4/24 Minimax lower bounds: Le Cam W15.1
4/29 Minimax lower bounds: proof of Le Cam W15.2
5/1 Minimax lower bounds using Fano W15.3
5/6 Proof of Fano's method W15.3.2, papers/baraudFano.pdf
5/8 Proof of Fano's method using information theory W15.3
5/13 Review
5/15 What we did not cover