Includes bibliographical references and index.

Essential Statistics for Data Science is a very short crash course for students entering a serious graduate program in data science without knowing enough statistics. However, it is not the type of introductory course that simply teaches students how to plug numbers into a formula and perform a t-test. While the course does start from the basics of probability and random variables, it moves along rapidly and ambitiously takes students in a matter of weeks to a number of relatively advanced topics in both frequentist and Bayesian inference as well as uncertainty assessment-such as the EM algorithm, the Gibbs sampler, and the bootstrap. The "main plot" unfolds in three parts. Part I, Talking Probability: The statistical approach to analysing data begins with a probability model to describe the data generating process; that's why, to study statistics, one must first learn to speak the language of probability. Part II, Doing Statistics: Before a model becomes truly useful, one must learn something about the unknown quantities in it-e.g., its parameters-from the data it is presumed to have generated, whether one cares about the parameters themselves or not; that's what much of statistical inference is about. Part III, Facing Uncertainty: Although one usually does not care much about parameters that don't have intrinsic scientific meaning, for those that do, it is important to explicitly describe how much uncertainty we have about.