Weekly update for week 36

Dear all,

I hope this week started the best possible way. Here coms the weekly digest  for FYS-STK3155/4155. 

Last week we discussed in depth the extensions of the standard linear regression called the Ridge and the Lasso methods. We discussed also the singular value decomposition algorithm and how can be used to analyse the results from the ridge and lasso methods.

Most of the material is covered by chapter 3.1-3.4 in the textbook by Hastie et al. My own lectures slides will be updated later today and should be ready by tomorrow (Wednesday).   Homework 2 is an extension of exercise 2 from last week and deals now with an implementation and analysis of the Ridge and the Lasso methods. See 

https://compphysics.github.io/MachineLearning/doc/web/course.html and scroll down to homework 2. Or just go to the files at

https://github.com/CompPhysics/MachineLearning/tree/master/doc/Projects/2018/hw2

This week we will wrap up this discussion with codes and further examples. Before venturing into classification schemes and logistic regression (next week), we will look at practical techniques like  various resampling methods (cross-validation methods and the famous bootstrap method chapters 7.9-7.10 of Hastie et al). 

We will also discuss dimension reduction methods and methods for higher dimensions, partly covered by chapter 7.  We will also start looking at more advanced examples with data including more predictors. Till now we have limited ourselves to just a simple function $$y(x)$$. 

At the lab we will thus work on homework 2 and this prepares you for parts of project 1. The latter will be available next week. 

Best wishes to you all,

Morten