CHE223H1

Statistics

Provides students with an introduction to statistical learning, namely the building of models from data. The course begins with foundational topics in elementary statistics. In the statistical learning portion of the course, the problem is formulated in terms of a system having input and output variables, the main goals of prediction and inference are presented, mean square error is defined, and the bias-variance trade-off is described in the context of overfitting the data. Statistical learning methodologies that are covered include K-Nearest Neighbours (KNN) regression, simple linear regression, multiple linear regression, and principal component analysis. Cross-validation is introduced as a popular method for model assessment and selection. The tutorial involves extensive computer-based simulation work to help students understand and appreciate the key concepts and to gain experience applying statistical learning to real data.