15.077/ESD.753  Statist Learning & Data Mining

Project Presentations on May 17

Dear 15.077 Class,

Professor Rajapakse and I have decided to devote a part of the last class (May 17) to presentations on your term projects. This is voluntary and the incentive is that you can incorporate the incredibly insightful comments of your classmates, TAs, and Professors into your final report (automatic extension granted). Please let me know if you are interested.

Thanks,

Roy Welsch

Announced on 09 May 2006  2:16  p.m. by Roy Welsch

Project Proposals

Dear Class,

If you have submitted a proposal and not heard from me, it means that I did not have any problem with your proposal. I will be contacting some of you with ideas that might be of help.

Sorry to keep you so busy when the weather is so nice.

Roy Welsch

Announced on 06 May 2006  7:27  p.m. by Roy Welsch

Additional change to #6A

In my previous correction, I forgot to specify that S should be a symmetric projection matrix. Note that H is symmetric.

Sorry and thanks to those with sharp pencils.

Announced on 04 May 2006  4:54  p.m. by Roy Welsch

Correction to test problem #6A

The second part of #6A should read as follows:

Also show that the diagonals of a projection matrix S (S^2=S) obey the inequality 0<=sii<=1. Consdider the HAT matrix is a special case (yhat =Hy). When the X matrix in the linear regression model contains a column of ones show that 1/n<=hii<=1. Construct a data set and a corresponding regression model in which the HAT diagonals are all 1.0. Is it true that if for the vector of 1's, S1=1, then 1/n<=sii?

Announced on 04 May 2006  6:04  a.m. by Roy Welsch

HW#3 and Project Info

Dear Class,

HW#3 (last one) has been posted along with all data sets. Term project details have also been posted.

Roy Welsch

Announced on 27 April 2006  9:09  p.m. by Roy Welsch