prc - STAT 760 October 3 2011 Lecture on principal...

• Notes
• 3

This preview shows pages 1–2. Sign up to view the full content.

STAT 760, October 3, 2011 Lecture on principal components Setup: See text about perceived utility of dimension reduction schemes. They are widely used, though sometimes difficult to interpret, and not so easily motivated from any first-principles on the dynamics or statistical fluctuations, and Prof Newton rarely uses them! There are different ways to tell the story about principal components. Roughly, they extend to multiple dimensions the following observation (show a 2-d scatterplot of points, centered): that onto any line (through the origin) points in the sample can be projected (perpendicularly, as opposed to in linear regression), and that one can consider the difference vector d i = x i - ˆ x i between each point and its projection, and the total error i || d i || 2 , as a function of the line. It turns out that the line which minimizes the error in this approximation also maximizes the variability of the projected points, when these points are viewed in the topology of the line. Population perpective: We start the story from the perspective of population principal compo- nents , as many accounts do (e.g. our text, Section 7.2). Let X be a random vector in R p which, wlog, has mean 0. Our focus is on its covariance matrix Σ and with constructing linear combina- tions that themselves are uncorrelated and explain ever increasing amounts of the overall variation. Specifically, if we take a vector v R p and define the new random variable ξ = v t X , then ξ has mean 0 and variance v t Σ v . We consider only unit vectors v t v

This preview has intentionally blurred sections. Sign up to view the full version.

This is the end of the preview. Sign up to access the rest of the document.
• Fall '08
• Staff
• Linear Algebra, Singular value decomposition, ﬁrst eigenvector v1

{[ snackBarMessage ]}

What students are saying

• As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

Kiran Temple University Fox School of Business ‘17, Course Hero Intern

• I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

Dana University of Pennsylvania ‘17, Course Hero Intern

• The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

Jill Tulane University ‘16, Course Hero Intern