Categorical Data Analysis

4 Credits.

This course introduces statistical analysis and supervised machine learning of data with categorical dependent variables. The course aims to apply statistical analysis and learning methods to social science research. The course starts with an intuitive understanding of the probability distributions and assumptions. It proceeds with applying estimators, model selections, and inferences to an example of social science research. It continues to address overfitting using LASSO and data complexity using tree classification, random forest, support vector machine (SVM), k-nearest neighbors (KNN), and neural networks.Categorical dependent variables can be binary, ordered-category, multi-category, and count (logit/probit, ordered-logit, multinomial logic, Poisson, negative binomial, loglinear and their corresponding lasso models). High-dimensional big data from sources including digital, administrative, complex surveys, and sensor data that provide covariates can be handled with supervised ML techniques. These models, methods, and techniques are accompanied with demonstrative instructions of empirical data analysis, including Stata code and substantive interpretations.