search

asadoughi / stat-learning

Notes and exercise attempts for "An Introduction to Statistical Learning"
http://asadoughi.github.io/stat-learning
2.13k stars 1.62k forks source link

Add solution for Chapter 6, Exercise 7, closes #27 #36

Closed jdavis closed 10 years ago

jdavis commented 10 years ago

I cleaned up my solution and added it. Let me know what you think! :smile:

jdavis commented 10 years ago

Any update on this?

asadoughi commented 10 years ago

Thanks @jdavis for your PR. In your answers to (c) and (e), there isn't a reference to the question's argument for mode and mean. Is there something I am missing?

jdavis commented 10 years ago

Ahh, yeah. I'll update it to be more explicit.

The question of asking for the mode for \beta is the same as asking "Argue that the LASSO/Ridge Regression is the most likely value for \beta under this posterior distribution."

In other words, with those selected distributions and using Bayesian probability, we should end up with the exact equations of the selected technique (Lasso/Ridge).

jdavis commented 10 years ago

@asadoughi Okay, I just updated it with more info on how the solution is in fact the mode for the Lasso solution and the mode/mean for the Ridge solution.

Here's two screenshots straight from the book to assure you that I'm not making it up :smile: : screen shot 2014-05-22 at 6 44 42 pm screen shot 2014-05-22 at 6 44 48 pm

jdavis commented 10 years ago

Is that better to understand?

asadoughi commented 10 years ago

Thanks for clarifying!