Designed to be taken concurrently with or after Principles and Techniques of Data Science (Data C100) or Probability for Data Science (Data C140) or both, each connector course consists of an intensive study of data science ideas in a particular field. Topics include the development of the theory of data science and the application of data science in a variety of domains. Topics vary by field, and more than one topic may be offered in a semester.

In this seminar we will discuss the theoretical basis for methods of statistical inference covered in Data 8, Data 100, and Data 140, as well as optimality properties and generalization. Topics will include modes of convergence of random variables, consistency, maximum likelihood and Fisher information, Kullback-Leibler divergence, empirical distributions and the bootstrap, the Neyman-Pearson paradigm, the t and F distributions via beta-gamma algebra, permutation-based methods, Kolmogorov-Smirnov tests, multiple comparisons, multiple testing, and the Bayesian viewpoint. Prerequisites are Data 100 and Data 140 (no exceptions). In addition, multivariable calculus is preferred. The seminar will be fast-paced and math-intensive. Students will be expected to learn independently and participate actively in all seminar and discussion sessions.

Repeat Rules