Intro to PCA
Adapted from G. Piatetsky-Shapiro, Biologically Inspired Intelligent Systems (Lecture 7) and R. Gutierrez-Osunas Lecture
Field Reduction Improves Classification
Most mining algorithms look for non-linear combinations of fields -- can easily find many spurious combinations given small # of records and large # of fields Classification accuracy improves if we first reduce number of fields Multi-class heuristic: select equal # of fields from each class
8
Selecting Most Relevant Fields
If there are too many fields, select a subset that is most relevant
Can select top N fields using 1-field predictive accuracy as computed earlier What is good N?
Rule of thumb: keep top 50 fields
Other techniques exist
Attribute Construction
Better to have a fair modeling method and good variables, than to have the best modeling method and poor variables Examples:
People are eligible for pension withdrawal at age 59 . Create it as a separate Boolean variable! Household income as sum of spouses incomes in loan underwritting
Advanced methods exists for automatically examining variable combinations, but it is very computationally expensive!
10
Variance
A measure of the spread of the data in a data set
X
s !
2 i! 1
X
n 1
Variance is claimed to be the original statistical measure of spread of data.
11
Covariance
Variance measure of the deviation from the mean for points in one dimension, e.g., heights Covariance a measure of how much each of the dimensions varies from the mean with respect to each other. Covariance is measured between 2 dimensions to see if there is a relationship between the 2 dimensions, e.g., number of hours studied & grade obtained. The covariance between one dimension and itself is the variance
12
Covariance
X
var( X ) !
i! 1 n n i
X
X
n 1
X
cov( X ,Y ) !
i! 1
X Yi Y
n 1
So, if you had a 3-dimensional data set (x,y,z), then you could measure the covariance between the x and y dimensions, the y and z dimensions, and the x and z dimensions.
13
Covariance
What is the interpretation of covariance calculations? Say you have a 2-dimensional data set
X: number of hours studied for a subject Y: marks obtained in that subject
And assume the covariance value (between X and Y) is: 104.53 What does this value mean?
14
Covariance
Exact value is not as important as its sign. A positive value of covariance indicates that both dimensions increase or decrease together, e.g., as the number of hours studied increases, the grades in that subject also increase. A negative value indicates while one increases the other decreases, or vice-versa, e.g., active social life at BYU vs. performance in CS Dept. If covariance is zero: the two dimensions are independent of each other, e.g., heights of students vs. grades obtained in a subject. 15
Covariance
Why bother with calculating (expensive) covariance when we could just plot the 2 values to see their relationship?
Covariance calculations are used to find relationships between dimensions in high dimensional data sets (usually greater than 3) where visualization is difficult.
16
Covariance Matrix
Representing covariance among dimensions as a matrix, e.g., for 3 dimensions:
cov( X, X) cov( X,Y ) cov( X,Z) C ! cov(Y , X) cov(Y ,Y ) cov(Y ,Z) cov(Z, X) cov(Z,Y ) cov(Z,Z)
Properties:
Diagonal: variances of the variables cov(X,Y)=cov(Y,X), hence matrix is symmetrical about the diagonal (upper triangular) n-dimensional data will result in nxn covariance matrix 17
Transformation Matrices
Consider the following:
3 3 3 12 2 v ! ! 4 v 2 2 2 1 8
The square (transformation) matrix scales (3,2) Now assume we take a multiple of (3,2) 3 6 2 v ! 2 4
2 2
3 v 1
6 ! 4
6 24 ! 4 v 4 16
18
Transformation Matrices
Scale vector (3,2) by a value 2 to get (6,4) Multiply by the square transformation matrix And we see that the result is still scaled by 4. WHY? A vector consists of both length and direction. Scaling a vector only changes its length and not its direction. This is an important observation in the transformation of matrices leading to formation of eigenvectors and eigenvalues. Irrespective of how much we scale (3,2) by, the solution (under the given transformation matrix) is always a multiple of 4.
19
Eigenvalue Problem
The eigenvalue problem is any problem having the following form: A.v= .v A: n x n matrix v: n x 1 non-zero vector : scalar Any value of for which this equation has a solution is called the eigenvalue of A and the vector v which corresponds to this value is called the eigenvector of A.
20
Eigenvalue Problem
Going back to our example:
3 3 3 12 2 v ! ! 4 v 2 2 2 1 8
A . v =
. v
Therefore, (3,2) is an eigenvector of the square matrix A and 4 is an eigenvalue of A The question is: Given matrix A, how can we calculate the eigenvector and eigenvalues for A?
21
Calculating Eigenvectors & Eigenvalues
Simple matrix algebra shows that: A.v= .v A.v- .I.v=0 (A - . I ). v = 0 Finding the roots of |A - . I| will give the eigenvalues and for each of these eigenvalues there will be an eigenvector Example
22
Calculating Eigenvectors & Eigenvalues
Let
0 1 A ! 2 3 Then:
0 1 0 0 1 1 P A P .I ! ! P 0 0 2 3 1 2 3
0 P
P 1 ! ! v 3 P
2 v 1
! P2 3P 2 P 2 3 P
And setting the determinant to 0, we obtain 2 eigenvalues: 1 = -1 and 2 = -2
23
Calculating Eigenvectors & Eigenvalues
For
1
the eigenvector is:
1 v1:1 . ! 0 v 2 1:2 and 2v1:1 2v1:2 ! 0
A P1 .I
.v1 ! 0
1 2
v1:1 v 1:2 ! 0 v1:1 ! v1:2
Therefore the first eigenvector is any column vector in which the two elements have equal magnitude and opposite sign.
24
Calculating Eigenvectors & Eigenvalues
Therefore eigenvector v1 is
1 v1 ! k1 1
where k1 is some constant. Similarly we find that eigenvector v2
1 v 2 ! k2 2
where k2 is some constant.
25
Properties of Eigenvectors and Eigenvalues
Eigenvectors can only be found for square matrices and not every square matrix has eigenvectors. Given an n x n matrix (with eigenvectors), we can find n eigenvectors. All eigenvectors of a symmetric* matrix are perpendicular to each other, no matter how many dimensions we have. In practice eigenvectors are normalized to have unit length.
*Note: covariance matrices are symmetric! 26
PCA
Principal components analysis (PCA) is a technique that can be used to simplify a dataset It is a linear transformation that chooses a new coordinate system for the data set such that
The greatest variance by any projection of the data set comes to lie on the first axis (then called the first principal component) The second greatest variance on the second axis Etc.
PCA can be used for reducing dimensionality by eliminating the later principal components.
27
PCA
By finding the eigenvalues and eigenvectors of the covariance matrix, we find that the eigenvectors with the largest eigenvalues correspond to the dimensions that have the strongest correlation in the dataset. These are the principal components. PCA is a useful statistical technique that has found application in:
Fields such as face recognition and image compression Finding patterns in data of high dimension.
28
PCA Process STEP 1
Subtract the mean from each of the dimensions This produces a data set whose mean is zero. Subtracting the mean makes variance and covariance calculation easier by simplifying their equations. The variance and co-variance values are not affected by the mean value.
29
PCA Process STEP 1
X Y X d 0.69 0.39 X !1.81 Y !1.91 0.09 1.29 0.49 0.19 Y d 0.49 0.99 0.29 1.09 0.79 0.31 2.5 2.4 0.5 0.7 2.2 2.9 1.9 2.2 3.1 3.0 2.3 2.7 2.0 1.6 1.0 1.1 1.5 1.6 1.2 0.9 1.31 1.21
0.81 0.81 0.31 0.31 0.71 1.01
30
http://kybele.psych.cornell.edu/~edelman/Psych-465-Spring-2003/PCA-tutorial.pdf
PCA Process STEP 2
Calculate the covariance matrix
0.616555556 0.615444444 cov ! 0.615444444 0.716555556
Since the non-diagonal elements in this covariance matrix are positive, we should expect that both the X and Y variables increase together. Since it is symmetric, we expect the eigenvectors to be orthogonal.
31
PCA Process STEP 3
Calculate the eigenvectors and eigenvalues of the covariance matrix 0.490833989 eigenvalues ! 1.28402771
0.735178656 0.677873399 eigenvectors ! 0.677873399 0.735178656
32
PCA Process STEP 3
Eigenvectors are plotted as diagonal dotted lines on the plot. (note: they are perpendicular to each other). One of the eigenvectors goes through the middle of the points, like drawing a line of best fit. The second eigenvector gives us the other, less important, pattern in the data, that all the points follow the main line, but are off to the side of the main line by some amount.
33
PCA Process STEP 4
Reduce dimensionality and form feature vector The eigenvector with the highest eigenvalue is the principal component of the data set. In our example, the eigenvector with the largest eigenvalue is the one that points down the middle of the data. Once eigenvectors are found from the covariance matrix, the next step is to order them by eigenvalue, highest to lowest. This gives the components in order of significance.
34
PCA Process STEP 4
Now, if youd like, you can decide to ignore the components of lesser significance. You do lose some information, but if the eigenvalues are small, you dont lose much
n dimensions in your data calculate n eigenvectors and eigenvalues choose only the first p eigenvectors final data set has only p dimensions.
35
PCA Process STEP 4
When the is are sorted in descending order, the proportion of variance explained by the p principal components is:
P P
i! 1 i! 1 n
!
i
P1 P 2 K P p P1 P 2 K P p K P n
If the dimensions are highly correlated, there will be a small number of eigenvectors with large eigenvalues and p will be much smaller than n. If the dimensions are not correlated, p will be as large as n and PCA does not help. 36
PCA Process STEP 4
Feature Vector FeatureVector = (
1 2 3 p)
(take the eigenvectors to keep from the ordered list of eigenvectors, and form a matrix with these eigenvectors in the columns)
We can either form a feature vector with both of the eigenvectors: 0.677873399 0.735178656
0.735178656 0.677873399
or, we can choose to leave out the smaller, less significant component and only have a single column:
0.677873399 0.735178656
37
PCA Process STEP 5
Derive the new data
FinalData = RowFeatureVector x RowZeroMeanData
RowFeatureVector is the matrix with the eigenvectors in the columns transposed so that the eigenvectors are now in the rows, with the most significant eigenvector at the top. RowZeroMeanData is the mean-adjusted data transposed, i.e., the data items are in each column, with each row holding a separate dimension.
38
PCA Process STEP 5
FinalData is the final data set, with data items in columns, and dimensions along rows. What does this give us? The original data solely in terms of the vectors we chose. We have changed our data from being in terms of the axes X and Y, to now be in terms of our 2 eigenvectors.
39
PCA Process STEP 5
FinalData (transpose: dimensions along columns)
ne X 0.827870186 1.77758033 0.992197494 0.274210416 1.67580142 0.912949103 0.0991094375 1.14457216 0.438046137 1.22382956 ne Y 0.175115307 0.142857227 0.384374989 0.130417207 0.209498461 0.175282444 0.349824698 0.0464172582 0.0177646297 0.162675287
40
PCA Process STEP 5
41
Reconstruction of Original Data
Recall that:
FinalData = RowFeatureVector x RowZeroMeanData
Then:
RowZeroMeanData = RowFeatureVector-1 x FinalData
And thus:
RowOriginalData = (RowFeatureVector-1 x FinalData) + OriginalMean
If we use unit eigenvectors, the inverse is the same as the transpose (hence, easier).
42
Reconstruction of Original Data
If we reduce the dimensionality (i.e., p<n), obviously, when reconstructing the data we lose those dimensions we chose to discard. In our example let us assume that we considered only a single eigenvector. The final data is newX only and the reconstruction yields
43
Reconstruction of original Data
The variation along the principal component is preserved. The variation along the other component has been lost.
44
45
A Word on Factor Analysis
The reciprocal of PCA(?) PCA generates new variables (zi) that are linear combinations of the original input variables (xi). FA assumes that there are factors (zi) that, when linearly combined, generate the input variables (xi).
46
A Word On Linear Discrimant Analysis
Both PCA and FA are unsupervised. LDA seeks to find a dimension such that when the data is projected onto it, the two classes* are well separated (i.e., the means are as far apart as possible and the examples of classes are as tightly clustered)
*This generalizes naturally to K classes yielding K-1 dimensions
47
References
PCA tutorial: http://kybele.psych.cornell.edu/~edelman/Ps ych-465-Spring-2003/PCA-tutorial.pdf Wikipedia: http://en.wikipedia.org/wiki/Principal_comp onent_analysis http://en.wikipedia.org/wiki/Eigenface
48
