An introduction to feature selection
Feature selection is the process of selecting a subset of relevant features (commonly known as predictors or independent variables) for model construction. Performing feature selection allows researchers to identify irrelevant data, improve the interpretation and increase predictive accuracy of learned models. A feature selection algorithm can be seen as the combination of a search technique for proposing new feature subsets, along with an evaluation which scores the different feature subsets. The choice of evaluation measure heavily influences the algorithm. There are three main categories of feature selection algorithms: wrappers, filters and embedded methods. In this seminar, we will introduce some basic feature selection methods such as score-based feature ranking, stepwise subset selection and LASSO regression.
Registration is free and open to all University of Waterloo faculty, staff, graduate and undergraduate students. The primary software we will discussed in this seminar is RStudio. There is no hands-on work in this seminar.