Cits5508-lect08-DimensionalityReduction

CITS5508 Machine Learning D´ ebora Corrˆ ea (Unit Coordinator and Lecturer) UWA 2023 1 / 36

Today Dimensionality Reduction and Principal Component Analysis. Hands-on Machine Learning with Scikit-Learn & TensorFlow (chapter 8) An Introduction to Statistical Learning (chapter 12) 2 / 36

Supervised and Unsupervised Learning In supervised learning: we have access to n features x 1 , x 2 , ..., x n measured on m observations. The goal is to predict an associate response variable ˆ y (that is, also measured on those m observations) using x 1 , x 2 , ..., x n . In unsupervised learning: we only have access to x 1 , x 2 , ..., x n measured on m observations. The goal is to discover patterns and interesting things about measurements on x 1 , x 2 , ..., x n . For instance: How can we visualise the data effectively? Can we find subgroups among the observations? Can we find subgroups among the features and use it for dimensionality reduction? 3 / 36