This is the 1st semester of a two-semester course that provides a rigorous introduction to methods and tools in advanced data analytics for social science doctoral students. The goal of the course is to provide students with a strong foundation of knowledge of core methods, thereby preparing them to contribute to research teams, to conduct their own research, and to enroll in more advanced courses. The course will cover research reproducibility (fall), machine learning (fall), natural language processing (spring), and causal inference (spring). In contrast to other courses currently offered on campus, this course’s intended audience is applied researchers, typically social science doctoral students in their 2nd or 3rd yr of graduate school.

This two-semester course provides a rigorous introduction to methods and tools in advanced data analytics for social science doctoral students. The course is divided into modules, each including lectures, discussion of example research articles, lab exercises, and a group project involving Python programming and executed in Jupyter notebooks. Course objectives: Proficiency with tools for reproducibility of research; Conceptual understanding of machine learning methods, including strengths and weaknesses of various algorithms and their appropriate to different kinds of prediction and classification problems; Conceptual understanding of experimental design and the structural causal model framework; Conceptual understanding of causal inference problems, solutions, and methods for longitudinal settings; Conceptual understanding of methods for extracting data from text using natural language processing; Ability to apply these concepts and execute relevant methodologies on social science data in Python and correctly interpret results. Prerequisites: A year-long course in statistical methods for social science graduate students (or equivalent prior experience), including multivariate regression (linear and non-linear models), maximum likelihood estimation, and introductory causal inference (omitted variable bias, potential outcomes, average treatment effects, causal graphs). Students without a background in introductory Python programming should take the D-Lab Python Fundamentals Workshop series

Repeat Rules

Course is not repeatable for credit.