Principal Components

42
Principal Components

description

Principal Components . Karl Pearson. Principal Components (PC). Objective : Given a data matrix of dimensions nxp (p variables and n elements) try to represent these data by using r variables (r

Transcript of Principal Components

Page 1: Principal Components

Principal Components

Page 2: Principal Components

• Karl Pearson

Page 3: Principal Components
Page 4: Principal Components

Principal Components (PC)

• Objective: Given a data matrix of dimensions nxp (p variables and n elements) try to represent these data by using r variables (r<p) with minimum lost of information

Page 5: Principal Components

We want to find a new set of p variables, Z, which are linear combinations of the original X variable such that :

• r of them contains all the information • The remaining p-r variables are noise

Page 6: Principal Components

First interpretation of principal components Optimal Data Representation

Page 7: Principal Components

xi

a

zi

ri

Proyection of a point in direction a: minimize the squared distanceImplies maximizing the variance (assuming zero mean variables)

xiT

xi = riT ri+ zT

i zi

Page 8: Principal Components
Page 9: Principal Components
Page 10: Principal Components
Page 11: Principal Components

Optimal Prediction

Find a new variable zi =a’Xi which is optimal to predictThe value of Xi in each element .

In general, find r variables, zi =Ar Xi , which are optimal to forecast All Xi with the least squared error criterion

It is easy to see that the solution is that zi =a’Xi must have maximum variance

Second interpretation of PC:

Page 12: Principal Components

The line which minimizes the orthogonal distance provides the axes of the ellipsoid

Third interpretation of PC

Find the optimal direction to represent the data. Axe of the ellipsoid which contains the data

This is idea of Pearson orthogonal regression

Page 13: Principal Components
Page 14: Principal Components
Page 15: Principal Components
Page 16: Principal Components

Second component

Page 17: Principal Components
Page 18: Principal Components
Page 19: Principal Components
Page 20: Principal Components
Page 21: Principal Components

Properties of PC

Page 22: Principal Components
Page 23: Principal Components
Page 24: Principal Components
Page 25: Principal Components
Page 26: Principal Components
Page 27: Principal Components

Standardized PC

Page 28: Principal Components
Page 29: Principal Components

Example Inves

Page 30: Principal Components

Example Inves

Page 31: Principal Components
Page 32: Principal Components

Example Medifis

Page 33: Principal Components
Page 34: Principal Components
Page 35: Principal Components

Example mundodes

Page 36: Principal Components

Example Mundodes

Page 37: Principal Components

Example for image analysis

Page 38: Principal Components
Page 39: Principal Components

The analysis have been done with 16 images. PC allows that Instead of sending 16 matrices of N2 pixels

16 3 70,616

we send a vector 16x3 with the values of the components and a matrix 3xN2 with the values of the new variables. We save

If instead of 16 images we have 100 images we save 95%

Page 40: Principal Components
Page 41: Principal Components
Page 42: Principal Components