Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
This lecture covers the Kernel K-means algorithm, an iterative procedure involving cluster initialization, data point assignment to centroids, and cluster update steps until stability. It explores the influence of terms on clustering using the RBF kernel, interpreting the objective function, and the impact of data point distribution on clustering results. The lecture also discusses interpreting the solution, density versus number of points, and the effect of polynomial kernels on clustering boundaries. Various methods for handling missing data, encoding categorical values, and dealing with unbalanced datasets are presented, along with techniques like down-sampling and over-sampling. The importance of dataset selection, preprocessing, and visualization is emphasized, with examples from different domains like gene expression, cancer RNA-Seq, and online retail datasets.