This document summarizes a presentation on practical machine learning for database administrators. It discusses using machine learning to classify PL/SQL code as good or bad, classify database schemas, cluster SQL statements, and detect anomalies in database workloads. The presentation covers what machine learning is, why it can be useful for databases, and provides examples of applying machine learning to common DBA problems like code classification. It describes building a naive Bayes classification model in Oracle to classify PL/SQL code, including extracting text features, training and testing the model, and assessing performance.