Version: June 19, 2007

Notes for Applied Multivariate

Analysis with MATLAB
These notes were written for use within Quantitative Psychology
courses at the University of Illinois, Champaign. The expectation is
that for Psychology 406/7 (Statistical Methods I and II), the mate-
rial up through Section 0.1.12 be available to a student. For Mul-
tivariate Analysis (Psychology 594) and Covariance Structure and
Factor Models (Psychology 588), the remainder of the notes are rel-
evant, with particular emphasis on Singular Value Decomposition
(SVD) and Eigenvector/Eigenvalue Decomposition (Spectral Decom-
position).

1
Contents

0.1 Necessary Matrix Algebra Tools . . . . . . . . . . . . 5

0.1.1 Preliminaries . . . . . . . . . . . . . . . . . . 5
0.1.2 The Data Matrix . . . . . . . . . . . . . . . . 12
0.1.3 Inner Products . . . . . . . . . . . . . . . . . 13
0.1.4 Determinants . . . . . . . . . . . . . . . . . . 17
0.1.5 Linear Independence/Dependence of Vectors . 19
0.1.6 Matrix Inverses . . . . . . . . . . . . . . . . . 20
0.1.7 Matrices as Transformations . . . . . . . . . . 23
0.1.8 Matrix and Vector Orthogonality . . . . . . . 25
0.1.9 Matrix Rank . . . . . . . . . . . . . . . . . . 25
0.1.10 Using Matrices to Solve Equations . . . . . . 29
0.1.11 Quadratic Forms . . . . . . . . . . . . . . . . 30
0.1.12 Multiple Regression . . . . . . . . . . . . . . 31
0.2 Eigenvectors and Eigenvalues . . . . . . . . . . . . . 33
0.3 The Singular Value Decomposition of a Matrix . . . . 41
0.4 Common Multivariate Methods in Matrix Terms . . . 42
0.4.1 Principal Components . . . . . . . . . . . . . 42
0.4.2 Discriminant Analysis . . . . . . . . . . . . . 43
0.4.3 Canonical Correlation . . . . . . . . . . . . . 44
0.4.4 Algebraic Restrictions on Correlations . . . . . 46
0.4.5 The Biplot . . . . . . . . . . . . . . . . . . . 47

2
0.4.6 The Procrustes Problem . . . . . . . . . . . . 49
0.4.7 Matrix Rank Reduction . . . . . . . . . . . . 50
0.4.8 Torgerson Metric Multidimensional Scaling . . 50
0.4.9 A Guttman Multidimensional Scaling Result . 52
0.4.10 A Few General MATLAB Routines to Know
About . . . . . . . . . . . . . . . . . . . . . . 53

3
List of Figures

1 Two vectors plotted in two-dimensional space . . . . 15

2 Illustration of projecting one vector onto another . . . 16

4
0.1 Necessary Matrix Algebra Tools

The strategies of multivariate analysis tend to be confusing unless

specied compactly in matrix terms. Therefore, we will spend some
signicant amount of time on these topics because, in fact, most of
multivariate analysis falls out directly once we have these tools under
control. Remember the old Saturday Night Live skit with Hans and
Franz, listen to me now, and believe me later. I have a goal in mind
of where I would like you all to be at the point of understanding
and being able to work with what is called the Singular Value De-
composition (SVD) of a matrix, and to understand the matrix topics
that lead up to the SVD. Very much like teaching how to use some
word-processing program, where we need to learn all the various com-
mands and what they do, an introduction to the matrix tools can
seem a little disjointed. But just like word-processing comes together
more meaningfully when required to do your own manuscripts from
beginning to end, once we proceed into the techniques of multivariate
analysis per se, the wisdom of this preliminary matrix excursion will
be apparent.

0.1.1 Preliminaries

A matrix is merely an array of numbers; for example,

4 1 3 1



4 6 0 2


7 2 1 4
is a matrix. In general, we denote a matrix by an uppercase (capi-
tal) boldface letter such as A (or using a proofreader representation

5
on the blackboard, a capital letter with a wavy line underneath to
indicate boldface):

a11 a12 a13 a1V


a21 a22 a23 a2V

A=

... ... ... . . . ...




aU1 aU2 aU3 aUV
This matrix has U rows and V columns and is said to have order
U V . An arbitrary element auv refers to the element in the uth
row and v th column, with the row index always preceding the column
index (and therefore, we might use the notation of A = {auv }UV
to indicate the matrix A as well as its order).
A 1 1 matrix such as (4)11 is just an ordinary number, called a
scalar. So without loss of any generality, numbers are just matrices.
A vector is a matrix with a single row or column; we denote a column
vector by a lowercase boldface letter, e.g., x, y, z, and so on. The
vector



x1

x=

...



xU U1

is of order U 1; the column indices are typically omitted since there

is only one. A row vector is written as

x = (x1, . . . , xU )1U
with the prime indicating the transpose of x, i.e., the interchange of
row(s) and column(s). This transpose operation can be applied to
any matrix; for example,

6



1 1


A=
3 7


4 1 32

1 3 4
A =

1 7 1 23
If a matrix is square, dened by having the same number of rows
as columns, say U , and if the matrix and its transpose are equal, the
matrix is said to be symmetric. Thus, in A = {auv }UU , auv = avu
for all u and v. As an example,

1 4 3

A = A =

4 7 1



3 1 3
For a square matrix AUU , the elements auu, 1 u U , lie along
the main or principal diagonal. The sum of main diagonal entries
of a square matrix is called the trace; thus,

trace(AUU ) tr(A) = a11 + + aUU

A number of special matrices appear periodically in the notes to
follow. A U V matrix of all zeros is called a null matrix, and might
be denoted by

0 0

= . . . . ...
..




0 0

7
Similarly, we might at times need an U V matrix of all ones, say
E:



1 1
E=

... . . . ...




1 1
A diagonal matrix is square with zeros in all the o main-diagonal
positions:

a1 0

DUU =

... . . . ...



0 aU UU
Here, we again indicate the main diagonal entries with just one index
as a1, a2, . . . , aU . If all of the main diagonal entries in a diagonal
matrix are 1s, we have the identity matrix denoted by I:

1 0 0


0 1 0
I=
.
.. ... ... ...




0 0 1
To introduce some useful operations on matrices, suppose we have
two matrices A and B of the same U V order:

a11 a1V

A=

... . . . ...



aU1 aUV UV

b11 b1V

B=

... . . . ...



bU1 bUV UV

8
As a denition for equality of two matrices of the same order (and
for which it only makes sense to talk about equality), we have:
A = B if and only if auv = buv for all u and v.
Remember, the if and only if statement (sometimes abbreviated
as i) implies two conditions:
if A = B, then auv = buv for all u and v;
if auv = buv for all u and v, then A = B.
Any denition by its very nature implies an if and only if state-
ment.
To add two matrices together, they rst have to be of the same
order (referred to as conformal for addition); we then do the addition
component by component:

a11 + b11 a1V + b1V

A+B=

... ... ...



aU1 + bU1 aUV + bUV UV

To preform scalar multiplication of a matrix A by, say, a constant

c, we again do the multiplication component by component:



ca11 ca1V

a11 a1V




cA =

... . . . ...
=c

... . . . ...



caU1 caUV aU1 aUV
Thus, if one wished to dene the dierence of two matrices, we could
proceed rather obviously as follows:

A B A + (1)B = {auv buv }

One of the more important matrix operations is multiplication
where two matrices are said to be conformal for multiplication if the
9
number of rows in one matches the number of columns in the second.
For example, suppose A is U V and B is V W ; then, because the
number of columns in A matches the number of rows in B, we can

dene AB as CUW , where {cuw } = { Vk=1 auk bkw }. This process
might be referred to as row (of A) by column (of B) multiplication;
the following simple example should make this clear:

1 4

1 2 0 1


A32 =
3 1 , B24 =

;
1 0 1 4
1 0

AB = C34 =

1(1) + 4(1) 1(2) + 4(0) 1(0) + 4(1) 1(1) + 4(4)



3(1) + 1(1) 3(2) + 1(0) 3(0) + 1(1) 3(1) + 1(4) =


1(1) + 0(1) 1(2) + 0(0) 1(0) + 0(1) 1(1) + 0(4)

3 2 4 17


2 6 1 7


1 2 0 1
Some properties of matrix addition and multiplication follow, where
the matrices are assumed conformal for the operations given:
(A) matrix addition is commutative:

A+B=B+A
(B) matrix addition is associative:

A + (B + C) = (A + B) + C

10
 (C) matrix multiplication is right and left distributive over matrix
addition:
A(B + C) = AB + AC

(A + B)C = AC + BC
(D) matrix multiplication is associative:

A(BC) = (AB)C
In general, AB =  BA even if both products are dened. Thus,
multiplication is not commutative as the following simple example
shows:

0 1 1 1 0 1 1 1
A22 =
; B22 = ; AB = ; BA =
1 0 0 1 1 1 1 0
In the product AB, we say that B is premultiplied by A and A
is postmultiplied by B. Thus, if we pre- or postmultiply a matrix
by the identity, the same matrix is retrieved:

IUU AUV = AUV ; AUV IV V = AUV

If we premultiply A by a diagonal matrix D, then each row of A is
multiplied by a particular diagonal entry in D:



d1a11 d1a1V

DUU AUV =

... ... ...



dU ar1 dU aUV
If A is post-multiplied by a diagonal matrix D, then each column of
A is multiplied by a particular diagonal entry in D:
11

d1a11 dV a1V

AUV DV V =

... ... ...



d1aU1 dV aUV
Finally, we end this section with a few useful results on the transpose
operation and matrix multiplication and addition:

(AB) = BA ; (ABC) = CBA; . . .

(A) = A; (A + B) = A + B

0.1.2 The Data Matrix

A very common type of matrix encountered in multivariate analysis

is what is referred to as a data matrix containing, say, observations
for N subjects on P variables. We will typically denote this matrix
by XNP = {xij }, with a generic element of xij referring to the
observation for subject or row i on variable or column j (1 i N
and 1 j P ):

x11 x12 x1P


x21 x22 x2P

XNP =

.
.. ... . . . ...




xN1 xN2 xNP
All right-thinking people always list subjects as rows and variables
as columns, conforming also to the now-common convention for com-
puter spreadsheets.
Any matrix in general, including a data matrix, can be viewed
either as a collection of its row vectors or of its column vectors,
12
and these interpretations can be generally useful. For a data matrix
XNP , let xi = (xi1, . . . , xiP )1P denote the row vector for subject
i, 1 i N , and let vj denote the N 1 column vector for variable
j:

x1j

vj =

...



xNj N1

Thus, each subject could be viewed as providing a vector of coordi-

nates (1 P ) in P -dimensional variable space, where the P axes
correspond to the P variables; or each variable could be viewed as
providing a vector of coordinates (N 1) in subject space, where
the N axes correspond to the N subjects:


x1



x2
XNP
=
...



= v1 v2 vP


xN

0.1.3 Inner Products

The inner product (also called the dot or scalar product) of two
vectors , xU1 and yU1 , is dened as

y1
U

x y = (x1, . . . , xU ) ...





= xuyu
u=1
yU
Thus, the inner product of a vector with itself is merely the sum of

squares of the entries in the vector: x x = Uu=1 x2u . Also, because

13
an inner product is a scalar and must equal it own transpose (i.e.,
x y = (x y) = y x), we have the end result that
x y = y x
If there is an inner product, there should also be an outer product
dened as the U U matrices given by xy or as yx . As indicated
by the display equations below, xy is the transpose of yx :



x1



x1y1 x1yU


xy =

...
(y1, . . . , yN ) =

... . . . ...



xU xU y1 xU yU

y1 y1 x1 y1 xU

 .. . .
.
yx = . (x1, . . . , xU ) = . . . . .
.


yU yU x1 yU xU
A vector can be viewed as a geometrical vector in U dimensional
space. Thus, the two 2 1 vectors

3 4
x=
; y =
4 1
can be represented in the two-dimensional Figure 1 below, with the
entries in the vectors dening the coordinates of the endpoints of the
arrows.
The Euclidean distance between two vectors, x and y, is given
as:


U

 (xu yu )2 = (x y)(x y)
u=1
and the length of any vector is the Euclidean distance between the
vector
and the origin. Thus, in Figure 1, the
distance between x and
y is 10 with respective lengths of 5 and 17.
14
 

(3,4)


4

3 
 
....
.....
....


....
2 ....
....
....

...
...


...



...
(4,1) ...

 
...

 
...
1

..


..

(0,0)

1 2 3 4
Figure 1: Two vectors plotted in two-dimensional space

The cosine of the angle between the two vectors x and y if dened
by:
xy
cos() =  1/2  1/2
(x x) (y y)
Thus, in the gure we have

4
3 4
1 16
cos() = = = .776
5 17 5 17
The cosine value of .776 corresponds to an angle of 39.1 degrees or
.68 radians; these later values can be found with the inverse (or arc)
cosine function (on, say, a hand calculator, or using MATLAB as we
suggest in the next section).
When the means of the entries in x and y are zero (i.e., deviations
from means have been taken), then cos() is the correlation between
15
 x



 

 c2 b2


....
....
...
...
...


...
...
...
..
..
..
..


..
. y
2
a

dy

Figure 2: Illustration of projecting one vector onto another

the entries in the two vectors. Vectors at right angles have cos() = 0,
or alternatively, the correlation is zero.
Figure 2 shows two generic vectors, x and y, where without loss
of any real generality, y is drawn horizontally in the plane and x
is projected at a right angle onto the vector y resulting in a point
dened as a multiple d of the vector y. The formula for d that
we demonstrate below is based on the Pythagorean theorem that
c2 = b2 + a2:

c2 = b2 + a2 x x = (x dy) (x dy) + d2y y

x x = x x dx y dy x + d2y y + d2y y

0 = 2dxy + 2d2y y

16
 x y
d= 
yy
The diagram in Figure 2 is somewhat constricted in the sense that the
angle between the vectors shown is less than 90 degrees; this allows
the constant d to be positive. Other angles might lead to negative d
when dening the projection of x onto y, and would merely indicate
the need to consider the vector y oriented in the opposite (negative)
direction. Similarly, the vector y is drawn with a larger length than
x which gives a value for d that is less than 1.0; otherwise, d would
be greater than 1.0 indicating a need to stretch y to represent the
point of projection onto it.
There are other formulas possible based on this geometric infor-
mation: the length of the projection is merely d times the length of
y; and cos() can be given as the length of dy divided by the length
  



of x, which is d y y/ x x = x y/( x x y y).

0.1.4 Determinants

To each square matrix, AUU , there is an associated scalar called the

determinant of A that is denoted by |A| or det(A). Determinants
up to a 3 3 can be given by formula:


a b
det( a 11) = a; det(
) = ad bc;
c d 22

a b c





det( d e f



) = aei + dhc + gf b (ceg + f ha + idb)

g h i 33

17
Beyond a 3 3 we can use a recursive process illustrated below. This
requires the introduction of a few additional matrix terms that we
now give: for a square matrix AUU , dene Auv to be the (n
1) (n 1) submatrix of A constructed by deleting the uth row
and v th column of A. We call det(Auv ) the minor of the entry auv ;
the signed minor of (1)u+v det(Auv ) is called the cofactor of auv .
The recursive algorithm would chose some row or column (rather
arbitrarily), and nd the cofactors for the entries in it; the cofactors
would then be weighted by the relevant entries and summed.
As an example, consider the 4 4 matrix

1 1 3 1

1 1


1 0


3 2 1 2


1 2 4 3
and choose the second row. The expression below involves the weighted
cofactors for 33 submatrices that can be obtained by formulas. Be-
yond a 4 4 there will be nesting of the processes:

1 3 1



1 3 1


2+1
(1)((1) ) det( 2 1 2 ) + (1)((1) ) det( 3 1 2



2+2
)+


2 4 3 1 4 3

1 1 1 1 1 3

2+3 2+4
(0)((1) ) det( 3 2 2

) + (1)((1) ) det( 3
2 1 ) =


1 2 3 1 2 4
5 + (15) + 0 + (29) = 39
Another strategy to nd the determinant of a matrix is to reduce it
a form in which we might note the determinant more or less by simple
18
inspection. The reductions could be carried out by operations that
have a known eect on the determinant; the form which we might
seek is a matrix that is either upper-triangular (all entries below the
main diagonal are all zero), lower-triangular (all entries above the
main diagonal are all zero), or diagonal. In these latter cases, the
determinant is merely the product of the diagonal elements. Once
found, we can note how the determinant might have been changed
by the reduction process and carry out the reverse changes to nd
the desired determinant.
The properties of determinants that we could rely on in the above
iterative process are as follows:
(A) if one row of A is multiplied by a constant c, the new determinant
is c det(A); the same is true for multiplying a column by c;
(B) if two rows or two columns of a matrix are interchanged, the sign
of the determinant is changed;
(C) if two rows or two columns of a matrix are equal, the determinant
is zero;
(D) the determinant is unchanged by adding a multiple of some row
to another row; the same is true for columns;
(E) a zero row or column implies a zero determinant;
(F) det(AB) = det(A) det(B)

0.1.5 Linear Independence/Dependence of Vectors

Suppose I have a collection of K vectors each of size U 1, x1, . . . , xK .

If no vector in the set can be written as a linear combination of the
remaining ones, the set of vectors is said to be linearly indepen-
dent ; otherwise, the vectors are linearly dependent. As an example,
19
consider the three vectors:

1 1 3


x1 = 4 ; x2 = 1 ; x3 = 7


0 1 1
Because 2x1 +x2 = x3 , we have a linear dependence among the three
vectors; however, x1 and x2, or, x2 and x3, are linearly independent.
If the U vectors (each of size U 1), x1, x2, . . . , xU , are linearly
independent, then the collection denes a basis, i.e., any vector can
be written as a linear combination of x1, x2, . . . , xU . For example,
using the standard basis, e1, e2, . . . , eU , where eu is a vector of all
zeros except for a single one in the uth position, any vector x =
(x1, . . . , xU ) can be written as:

x1 1 0 0


x2 0 1 0
= x +x + + x =
.
... ... ...
1 2 U
..


xU 0 0 1
x1 e 1 + x2 e 2 + + xU e U
Bases that consist of orthogonal vectors (where all inner products are
zero) are important later in what is known as principal components
analysis. The standard basis involves orthogonal vectors, and any
other basis may always be modied by what is called the Gram-
Schmidt orthogonalization process to produce a new basis that does
contain all orthogonal vectors.

0.1.6 Matrix Inverses

Suppose A and B are both square and of size U U . If AB = I,

then B is said to be an inverse of A and is denoted by A1( B).
20
Also, if AA1 = I, then A1A = I holds automatically. If A1
exists, the matrix A is said to be nonsingular ; if A1 does not
exist, A is singular.
An example:

1 3 1/5 3/5 1 0
=
2 1 2/5 1/5 0 1

1/5 3/5 1 3 1 0
=
2/5 1/5 2 1 0 1
Given a matrix A, the inverse A1 can be found using the follow-
ing four steps:
(A) form a matrix of the same size as A containing the minors for
all entries of A;
(B) multiply the matrix of minors by (1)u+v to produce the
matrix of cofactors;
(C) divide all entries in the cofactors matrix by det(A);
(D) the transpose of the matrix found in (C) gives A1.
As a mnemonic device to remember these four steps, we have the
phrase My Cat Does Tricks for Minor, Cofactor, Determinant Di-
vision, Transpose (I tried to work my cat turns tricks into the
appropriate phrase but failed with the second to the last t). Ob-
viously, an inverse exists for a matrix A if det(A) = 0, allowing the
division in step (C) to take place.
An example: for



1 3 2

A=
0 1 1 ; det(A) = 1


0 2 1
21
 Step (A), the matrix of minors:



1 0 0



1 1 2



1 1 1
Step (B), the matrix of cofactors:



1 0 0


1 1 2



1 1 1
Step (C), determinant division:

1 0 0



1 1 2



1 1 1
Step (D), matrix transpose:

1 1 1

A1 =
0 1
1



0 2 1
We can easily verify that AA1 = I:

1 3 2 1 1 1 1 0 0



0 1 1

0 1 1




= 0 1 0


0 2 1 0 2 1 0 0 1
As a very simple instance of the mnemonic in the case of a 2 2
matrix with arbitrary entries:

a b
A=

c d
22
the inverse exists if det(A) = ad bc = 0:

1 d b
A1 =

ad bc c a
Several properties of inverses are given below that will prove useful
in our continuing presentation:
(A) if A is symmetric, then so is A1 ;
(B) (A )1 = (A1); or, the inverse of a transpose is the transpose
of the inverse;
(C) (AB)1 = B1A1 ; (ABC)1 = C1B1A1 ; or, the in-
verse of a product is the product of inverses in the opposite order;
(D) (cA)1 = ( 1c )A1; or, the inverse of a scalar times a matrix
is the scalar inverse times the matrix inverse;
(E) the inverse of a diagonal matrix, is also diagonal with the
entries being the inverses of the entries from the original matrix (as-
suming none are zero):
1
a1 0
1
a1 0

..
.
. . . ...
=

... . . . ...



0 aU 0 1
aU

0.1.7 Matrices as Transformations

Any U V matrix A can be seen as transforming a V 1 vector

xV 1 to another U 1 vector yU1 :

yU1 = AUV xV 1
or,

23



y1



a11 a1V

x1



...
=

... . . . ...

...



yU aU1 aUV xV
where yu = au1x1 + au2x2 + + auV xV . Alternatively, y can be
written as a linear combination of the columns of A with weights
given by x1, . . . , xV :

y1 a11 a12 a1V

.. . ... ...
. = x1 .. + x2 + + xV


yU aU1 aU2 aUV
To indicate one common usage for matrix transformation in a data
context, suppose we consider our data matrix X = {xij }NP , where
xij represents an observation for subject i on variable j. We would
like to use matrix transformations to produce a standardized matrix
Z = {(xij xj )/sj }NP , where xj is the mean of the entries in
the j th column and sj is the corresponding standard deviation; thus,
the columns of Z all have mean zero and standard deviation one. A
matrix expression for this transformation could be written as follows:
1
ZNP = (INN ( )ENN )XNP DP P
N
where I is the identity matrix, E contains all ones, and D is a diagonal
matrix containing s11 , s12 , . . . , s1 , along the main diagonal positions.
P
Thus, (INN ( N )ENN )XNP produces a matrix with columns
1

deviated from the column means; a postmultiplication by D carries

out the within column division by the standard deviations. Finally, if
we dene the expression ( N1 )(ZZ)P P RP P , we have the familiar
correlation coecient matrix among the P variables.
24
0.1.8 Matrix and Vector Orthogonality

Two vectors, x and y, are said to be orthogonal if x y = 0, and

would lie at right angles when
graphed. If, in addition, x and y are

both of unit length (i.e., x x = y y = 1), then they are said to
be orthonormal. A square matrix TUU is said to be orthogonal
if its rows form a set of mutually orthonormal vectors. An example
(called a Helmert matrix of order 3) follows:



1/3 1/3 1/ 3

T= 1/ 2 1/ 2 0



1/ 6 1/ 6 2/ 6
There are several nice properties of orthogonal matrices that we
will see again in our various discussions to follow:
(A) TT = T T = I;
(B) the columns of T are orthonormal;
(C) det(T) = 1;
(D) if T and R are orthogonal, then so is TR;
(E) vectors lengths do not change under an orthogonal transfor-
mation: to see this, let y = Tx; then

y y = (Tx) (Tx) = x T Tx = x Ix = x x

0.1.9 Matrix Rank

An arbitrary matrix, A, of order U V can be written either in terms

of its U rows, say, r1, r2, . . . , rU or its V columns, c1, c2, . . . , cV ,
where

25


a1v

ru = au1 auV ; cv =

...



aUv
and

r1


r2

AUV =

...


= c1 c2 cV


rU
The maximum number of linearly independent rows of A and the
maximum number of linearly independent columns is the same; this
common number is dened to be the rank of A. A matrix is said to
be of full rank is the rank is equal to the minimum of U and V .
Matrix rank has a number of useful properties:
(A) A and A have the same rank;
(B) A A, AA, and A have the same rank;
(C) the rank of a matrix is unchanged by a pre- or postmultipli-
cation by a nonsingular matrix;
(D) the rank of a matrix is unchanged by what are called elemen-
tary row and column operations: (a) interchange of two rows or two
columns; (2) multiplication or a row or a column by a scalar; (3) ad-
dition of a row (or column) to another row (or column). This is true
because any elementary operation can be represented by a premul-
tiplication (if the operation is to be on rows) or a postmultiplication
(if the operation is to be on columns) of a nonsingular matrix.
To give a simple example, suppose we wish to perform some ele-
mentary row and column operations on the matrix
26

1 1 1



1 0 2


3 2 4
To interchange the rst two rows of this latter matrix, interchange the
rst two rows of an identity matrix and premultiply; for the rst two
columns to be interchanged, carry out the operation on the identity
and post-multiply:



0 1 0 1 1 1

1 0 2



1 0 0






1 0 2 = 1 1 1


0 0 1 3 2 4 3 2 4



1 1 1 0 1 0

1 1 1



1 0 2 1 0 0 = 0 1 2


3 2 4 0 0 1 2 3 4
To multiply a row of our example matrix (e.g., the second row by 5),
multiply the desired row of an identity matrix and premultiply; for
multiplying a specic column (e.g., the second column by 5), carry
out the operation of the identity and post-multiply:



1 0 0
1 1 1

1 1 1



0 5 0





1 0 2 = 5 0 10


0 0 1 3 2 4 3 2 4



1 1 1 1 0 0

1 5 1


1 0 2 0 5 0 = 1 0 2


3 2 4 0 0 1 3 10 4
27
To add one row to a second (e.g., the rst row to the second), carry
out the operation on the identity and premultiply; to add one column
to a second (e.g., the rst column to the second), carry out the
operation of the identity and post-multiply:

1 0 0 1 1 1 1 1 1


1 1 0 1 0 2 = 2 1 3


0 0 1 3 2 4 3 2 4



1 1 1 1 0 0

1 2 1



1 0 2






1 1 0 = 1 1 2


3 2 4 0 0 1 3 5 4
In general, by performing elementary row and column operations,
any U V matrix can be reduced to a canonical form:



1 0 0 0


... ... ... ... ... ...







0 1 0 0





0 0 0 0


... ... ... ... ... ...



0 0 0 0
The rank of a matrix can then be found by counting the number of
ones in the above matrix.
Given an U V matrix, A, there exist s nonsingular elementary
row operation matrices, R1 , . . . , Rs , and t nonsingular elementary
column operation matrices, C1, . . . , Ct such that Rs R1AC1 Ct
is in canonical form. Moreover, if A is square (U U ) and of full
rank (i.e., det(A) = 0), then there are s nonsingular elementary row

28
operation matrices, R1, . . . , Rs, and t nonsingular elementary col-
umn operation matrices, C1, . . . , Ct, such that Rs R1A = I or
AC1 Ct = I. Thus, A1 can be found either as Rs R1 or as
C1 Ct. In fact, a common way in which an inverse is calculated
by hand starts with both A and I on the same sheet of paper;
when reducing A step-by-step, the same operations are then applied
to I, building up the inverse until the canonical form is reached in
the reduction of A.

0.1.10 Using Matrices to Solve Equations

Suppose we have a set of U equations in V unknowns:

a11x1 + + a1V x1 = c1
... ... ... ... ... ...
aU1x1 + + aUV xV = cU
If we let

a11 a1V x1 c1

A=

... . . . ...
; x=

...
; c=

...



aU1 aUV xV cU
then the equations can be written as follows: AUV xV 1 = cU1 .
In the simplest instance, A is square and nonsingular, implying that
a solution may be given simply as x = A1c. If there are fewer (say,
S min(U, V ) linearly independent) equations than unknowns (so,
S is the rank of A), then we can solve for S unknowns in terms of the
constants c1, . . . , cU and the remaining V S unknowns. We will
see how this works in our discussion of obtaining eigenvectors that
correspond to certain eigenvalues in a section to follow. Generally,

29
the set of equations is said to be consistent if a solution exists, i.e., a
linear combination of the column vectors of A can be used to dene
c:

a11 a1V c1

x1

.
.
.
+ + xV
.
.
. .
= ..


aU1 aUV cU
or the augmented matrix (A c) has the same rank as A; otherwise no
solution exists and the system of equations is said to be inconsistent.

0.1.11 Quadratic Forms

Suppose AUU is symmetric and let x = (x1, . . . , xU ). A quadratic

form is dened by
U 
 U

x Ax = auv xu xv =
u=1 v=1

a11x21+a22x22+ +aUU x2U +2a12x1x2+ +2a1U x1xU + +2a(U1)U xU1 xU


For example, Uu=1(xu x)2, where x is the mean of the entries in x,
is a quadratic form since it can be written as

x1 (U 1)/U 1/U 1/U x1




x2





1/U (U 1)/U 1/U


x2



...


... ... ... ...
...



xU 1/U 1/U (U 1)/U xU
Because of the ubiquity of sum-of-squares in statistics, it should be
no surprise that quadratic forms play a central role in multivariate
analysis.
A symmetric matrix A (and associated quadratic form) are called
positive definite(p.d.) if xAx > 0 for all x = 0 (the zero vector);
30
if x Ax 0 for all x, then A is positive semi-definite(p.s.d). We
could have negative denite, negative semi-denite, and indenite
forms as well. Note that a correlation or covariance matrix is at
least positive semi-denite, and satises the stronger condition of
being positive denite if the vectors of the variables on which the
correlation or covariance matrix is based, are linearly independent.

0.1.12 Multiple Regression

One of the most common topics in any beginning statistics class is

multiple regression that we now formulate (in matrix terms) as the
relation between a dependent random variable Y and a collection
of K independent variables, X1, X2, . . . , XK . Suppose we have N
subjects on which we observe Y , and arrange these values into an
N 1 vector:

Y1


Y2
Y=
.
..





YN
The observations on the K independent variables are also placed in
vectors:

X11 X12 X1K


X21 X22 X2K

X1 = ; X = ; . . . ; X =
...
2
...
K
...



XN1 XN2 XNK
It would be simple if the vector Y were linearly dependent on X1, X2, . . . , XK
since then

31
 Y = b 1 X1 + b 2 X2 + + b K XK
for some values b1, . . . , bK . We could always write for any values of
b1 , . . . , bK :

Y = b 1 X1 + b 2 X2 + + b K XK + e
where
e1

e= .
..



eN
is an error vector. To formulate our task as an optimization problem
(least-squares), we wish to nd a good set of weights, b1, . . . , bK , so
the length of e is minimized, i.e., ee is made as small as possible.
As notation, let

YN1 = XNK bK1 + eN1

where

b1

X = X1 . . . XK ; b = .
..




bK
To minimize ee = (Y Xb)(Y Xb), we use the vector b that
satises what are called the normal equations:

X Xb = X Y
If XX is nonsingular (i.e., det(XX) = 0; or equivalently, X1, . . . , XK
are linearly independent), then
32
 b = (XX)1X Y
The vector that is closest to Y in our least-squares sense, is Xb;
this is a linear combination of the columns of X (or in other jargon,
Xb denes the projection of Y into the space dened by (all linear
combinations of) the columns of X.
In statistical uses of multiple regression, the estimated variance-
covariance matrix of the regression coecients, b1, . . . , bK , is given as
1
( NK )ee(XX)1, where ( NK1
)ee is an (unbiased) estimate of the
error variance for the distribution from which the errors are assumed
drawn. Also, in multiple regression instances that usually involve an
additive constant, the latter is obtained from a weight attached to
an independent variable dened to be identically one.
In multivariate multiple regression where there are, say, T depen-
dent variables (each represented by an N 1 vector), the dependent
vectors are merely concatenated together into an N T matrix,
YNT ; the solution to the normal equations now produces a matrix
BKT = (XX)1X Y of regression coecients. In eect, this gen-
eral expression just uses each of the dependent variables separately
and adjoins all the results.

0.2 Eigenvectors and Eigenvalues

Suppose we are given a square matrix, AUU , and consider the poly-
nomial det(A I) in the unknown value , referred to as Laplaces
expansion:
det(A I) = ()U + S1()U1 + + SU1 ()1 + SU ()0
33
where Su is the sum of all u u principal minor determinants. A
principal minor determinant is obtained from a submatrix formed
from A that has u diagonal elements left in it. Thus, S1 is the trace
of A and SU is the determinant.
There are U roots, 1, . . . , U , of the equation det(A I) = 0,
given that the left-hand-side is a U th degree polynomial. The roots
are called the eigenvalues of A. There are a number of properties
of eigenvalues that prove generally useful:
 
(A) det A = Uu=1 u ; trace(A) = Uu=1 u;
(B) if A is symmetric with real elements, then all u are real;
(C) if A is positive denite, then all u are positive (strictly greater
than zero); if A is positive semi-denite, then all u are nonnegative
(greater than or equal to zero);
(D) if A is symmetric and positive semi-denite with rank R, then
there are R positive roots and U R zero roots;
(E) the nonzero roots of AB are equal to those of BA; thus, the
trace of AB is equal to the trace of BA;
(F) eigenvalues of a diagonal matrix are the diagonal elements
themselves;
(G) for any U V matrix B, the ranks of B, BB, and BB
are all the same. Thus, because B B (and BB) are symmetric and
positive semi-denite (i.e., x(BB)x 0 because (Bx) (Bx) is a
sum-of-squares which is always nonnegative), we can use (D) to nd
the rank of B by counting the positive roots of BB.
We carry through a small example below:

34

7 0 1


A=
0 7 2


1 2 3

S1 = trace(A) = 17

7 0 7 1 7 2
S2 = det(
)+det( )+det( ) = 49+20+17 = 86
0 7 1 3 2 3
S3 = det(A) = 147 + 0 + 0 7 28 0 = 112
Thus,

det(A I) = ()3 + 17()2 + 86()1 + 112 =

3 + 172 86 + 112 = ( 2)( 8)( 7) = 0
which gives roots of 2, 8, and 7.
If u is an eigenvalue of A, then the equations [A uI]xu = 0
have a nontrivial solution (i.e., the determinant of A u I vanishes,
and so the inverse of A u I does not exist). The solution is called
an eigenvector (associated with the corresponding eigenvalue), and
can be characterized by the following condition:

Axu = uxu
An eigenvector is determined up to a scale factor only, so typically
we normalize to unit length (which then gives a option for the two
possible unit length solutions).
We continue our simple example and to nd the corresponding
eigenvalues: when = 2, we have the equations (for [A I]x = 0)
35



5 0 1
x1



0


0 5 2 x2

=
0


1 2 1 x3 0
with an arbitrary solution of



15 a



25 a


a
Choosing a to be + 530 to obtain one of the two possible normalized
solutions, we have as our nal eigenvector for = 2:

30
1


2
30


5
30

For = 7 we will use the normalized eigenvector of

5
2


1

5

0
and for = 8,


1
6
2




6
1

6

One of the interesting properties of eigenvalues/eigenvectors for a

symmetric matrix A is that if u and v are distinct eigenvalues,
36
then the corresponding eigenvectors, xu and xv , are orthogonal (i.e.,
xu xv = 0). We can show this in the following way: the dening
conditions of
Axu = uxu

Axv = v xv
lead to
xv Axu = xv uxu

xu Axv = xu v xv

Because A is symmetric and the left-hand-sides of these two expres-
sions are equal (they are one-by-one matrices and equal to their own
transposes), the right-hand-sides must also be equal. Thus,

xv uxu = xu v xv

xv xu u = xu xv v
Due to the equality of xv xu and xuxv , and by assumption, u = v ,
the inner product xv xu must be zero for the last displayed equality
to hold.
In summary of the above discussion, for every real symmetric ma-
trix AUU , there exists an orthogonal matrix P (i.e., P P = PP =
I) such that P AP = D, where D is a diagonal matrix containing
the eigenvalues of A, and

P = p1 . . . pU
37
where pu is a normalized eigenvector associated with u for 1 u
U . If the eigenvalues are not distinct, it is still possible to choose the
eigenvectors to be orthogonal. Finally, because P is an orthogonal
matrix (and P AP = D PP APP = PDP ), we can nally
represent A as

A = PDP
In terms of the small numerical example being used, we have for
P AP = D:


130 230 5
30 7 0 1
130 25 1
6



25 1 0
0 7 2

230 1 2
=
5 5 6

1 2 1 1 2 3 5 0 1
6 6 6 30 6

2 0 0


0 7 0


0 0 8
and for PDP = A:

130 25 1
6 2 0 0
130 230 5
30



230 1
5
2
6

0 7 0

25 1
5
0
=


5 0 1 0 0 8 1 2 1
30 6 6 6 6

7 0 1


0 7 2


1 2 3
The representation of A as PDP leads to several rather nice
computational tricks. First, if A is p.s.d., we can dene

38




1 . . . 0

D 1/2
.
.. . . . ...




0 . . . U
and represent A as

A = PD1/2D1/2P = PD1/2(PD1/2) = LL , say.

In other words, we have factored A into LL, for

1/2
L = PD = 1 p 1 2 p 2 . . . U p U
Secondly, if A is p.d., we can dene

1
1 ... 0

D 1


... . . . ...


1
0 ... U

and represent A1 as

A1 = PD1P
To verify,
AA1 = (PDP )(PD1P ) = I
Thirdly, to dene a square root matrix, let A1/2 PD1/2P. To
verify, A1/2A1/2 = PDP = A.
There is a generally interesting way to represent the multiplication
of two matrices considered as collections of column and row vectors,
respectively, where the nal answer is a sum of outer products of
vectors. This view will prove particularly useful in our discussion of

39
principal component analysis. Suppose we have two matrices BUV ,
represented as a collection of its V columns:

B = b1 b2 . . . bV
and CV W , represented as a collection of its V rows:


c1


c2
C=
.
..





cV
The product BC = D can be written as


c1


c
2
BC = b1 b2 . . . bV .
..



=


cV
b1c1 + b2c2 + + bV cV = D
As an example, consider the spectral decomposition of A consid-
ered above as PDP , and where from now on, without loss of any gen-
erality, the diagonal entries in D are ordered as 1 2 U .
We can represent A as



1 p 1

.

AUU = 1 p 1 . . . U p U .
. =



U p U
1p1p1 + + U pU pU
If A is p.s.d. and of rank R, then the above sum obviously stops at
R components. In general, the matrix BUU that is a rank K ( R)
40
least-squares approximation to A can be given by
B = 1p1p1 + + k pK pK
and the value of the loss function:
U 
 U 1
(auv buv )2 = (2K+1 + + 2U ) 2
v=1 u=1

0.3 The Singular Value Decomposition of a Matrix

The singular value decomposition (SVD) or the basic structure

of a matrix refers to the representation of any rectangular U V
matrix, say, A, as a triple product:

AUV = PUR RR QRV

where the R columns of P are orthonormal; the R rows of Q are
orthonormal; is diagonal with ordered positive entries, 1 2
R > 0; and R is the rank of A. Or, alternatively, we can ll
up this decomposition as

AUV = PUU UV QV V

where the columns of P and rows of Q are still orthonormal, and
the diagonal matrix forms the upper-left-corner of :


=


here, represents an appropriately dimensioned matrix of all zeros.
In analogy to the least-squares result of the last section, if a rank K
( R) matrix approximation to A is desired, say BUV , the rst K
ordered entries in are taken:
41
 B = 1p1q1 + + K pK qK
and the value of the loss function:
V 
 U
(auv buv )2 = K+1
2
+ + R2
v=1 u=1

This latter result of approximating one matrix (least-squares) by

another of lower rank, is referred to as the Ekart-Young theorem in
the psychometric literature.
Once one has the SVD of a matrix, a lot of representation needs
can be expressed in terms of it. For example, suppose A = PQ ;
the spectral decomposition of AA can then be given as
(PQ )(PQ ) = PQ QP = PP = P2P
Similarly, the spectral decomposition of A A is expressible as Q2 Q.

0.4 Common Multivariate Methods in Matrix Terms

In this section we give a very brief overview of some common meth-

ods of multivariate analysis in terms of the matrix ideas we have
introduced thus far in this chapter. Later chapters (if they ever get
writtten) will come back to these topics and develop them in more
detail.

0.4.1 Principal Components

Suppose we have a data matrix XNP = {xij }, with xij referring as

usual to the observation for subject i on variable or column j:
42

x11 x12 x1P


x21 x22 x2P

XNP =

.
.. ... . . . ...




xN1 xN2 xNP
The columns can be viewed as containing N observations on each of
P random variables that we denote generically by X1, X2, . . . , XP .
We let A denote the P P sample covariance matrix obtained among
the variables from X, and let 1 P 0 be its P eigenvalues
and p1, . . . , pP the corresponding normalized eigenvectors. Then,
the linear combination
X1


pk ...




XP
is called the k th (sample) principal component.
There are (at least) two interesting properties of principal compo-
nents to bring up at this time:
A) The k th principal component has maximum variance among
all linear combinations dened by unit length vectors orthogonal to
p1, . . . , pk1; also, it is uncorrelated with the components up to k1;
B) A 1p1p1 + + K pK pK gives a least-squares rank K
approximation to A (a special case of the Ekart-Young theorem for
an arbitrary symmetric matrix).

0.4.2 Discriminant Analysis

Suppose we have a one-way analysis-of-variance (ANOVA) layout

with J groups (nj subjects in group j, 1 j J), and P measure-
43
ments on each subject. If xijk denotes person i, in group j, and the
observation of variable k (1 i nj ; 1 j J; 1 k P ), then
dene the Between-Sum-of-Squares matrix
J

BP P = { xjk xk )(
nj ( xjk xk )}P P
j=1

and the Within-Sum-of-Squares matrix

nj
J 

WP P = { (xijk xjk )(xijk xjk )}P P
j=1 i=1

For the matrix product W1B, let 1, . . . , T 0 be the eigen-

vectors (T = min(P, J 1), and p1, . . . , pT the corresponding nor-
malized eigenvectors. Then, the linear combination



X1

pk

...



XP
is called the k th discriminant function. It has the valuable property
of maximizing the univariate F -ratio subject to being uncorrelated
with the earlier linear combinations. A variety of applications of
discriminant functions exists in classication that we will come back
to later. Also, standard multivariate ANOVA signicance testing is
based on various functions of the eigenvalues 1, . . . , T and their
derived sampling distributions.

0.4.3 Canonical Correlation

Suppose the collection of P random variables that we have observed

over the N subjects is actually in the form of two batteries, X1, . . . , XQ

44
and XQ+1, . . . , XP , and the observed covariance matrix AP P is par-
titioned into four parts:

A11 A12
AP P =



A12 A22
where A11 is Q Q and represents the observed covariances among
the variables in the rst battery; A22 is (P Q) (P Q) and
represents the observed covariances among the variables in the second
battery; A12 is Q (P Q) and represents the observed covariances
between the variables in the rst and second batteries. Consider the
following two equations in unknown vectors a and b, and unknown
scalar :
A1 1 
11 A12 A22 A12 a = a

A1  1
22 A12 A11 A12 b = b

There are T solutions to these expressions (for T = min(Q, (P

Q))), given by normalized unit-length vectors, a1 , . . . , aT and b1, . . . , bT ;
and a set of common 1 T 0.
The linear combinations of the rst and second batteries dened
by ak and bk are the k th canonical variates and have squared cor-
relation of k ; they are uncorrelated with all other canonical variates
(dened either in the rst or second batteries). Thus, a1 and b1
are the rst canonical variates with squared correlation of 1; among
all linear combinations dened by unit-length vectors for the vari-
ables in the two batteries, this squared correlation is the highest it
can be. (We note that the coecient matrices A1 1 
11 A12 A22 A12 and
A1  1
22 A12 A11 A12 are not symmetric; thus, special symmetrizing and

45
equivalent equation systems are typically used to obtain the solutions
to the original set of expressions.)

0.4.4 Algebraic Restrictions on Correlations

A matrix AP P that represents a covariance matrix among a collec-

tion of random variables, X1, . . . , XP is p.s.d.; and conversely, any
p.s.d. matrix represents the covariance matrix for some collection of
random variables. We partition A to isolate its last row and column
as

B(P 1)(P 1) g(P 1)1
A=



g aP P
B is the (P 1) (P 1) covariance matrix among the variables
X1, . . . , XP 1; g is (P 1) 1 and contains the cross-covariance
between the the rst P 1 variables and the P th; aP P is the variance
for the P th variable.
Based on the observation that determinants of p.s.d. matrices are
nonnegative, and a result on expressing determinants for partitioned
matrices (that we do not give here), it must be true that

gB1g aP P
or if we think correlations rather than merely covariances (so the
main diagonal of A consists of all ones):

g B1g 1
Given the correlation matrix B, the possible values the correlations
in g could have are in or on the ellipsoid dened in P 1 dimensions
by gB1g 1. The important point is that we do not have a box
46
in P 1 dimensions containing the correlations with sides extending
the whole range of 1; instead, some restrictions are placed on the
observable correlations that gets dened by the size of the correlation
in B. For example, when P = 3, a correlation between variables
X1 and X2 of r12 = 0 gives the degenerate ellipse of a circle for
constraining the correlation values between X1 and X2 and the third
variable X3 (in a two-dimensional r13 versus r23 coordinate system);
for r12 = 1, the ellipse attens to a line in this same two-dimensional
space.
Another algebraic restriction that can be seen immediately is based
on the formula for the partial correlation between two variables,
holding the third constant:
r12 r13r23

(1 r13
2 )(1 r 2 )
23

Bounding the above by 1 (because it is a correlation) and solving

for r12, gives the algebraic upper and lower bounds of

r12 r13r23 + (1 r13
2
)(1 r23
2
)

r13r23 (1 r13
2 )(1 r 2 ) r
23 12

0.4.5 The Biplot

Let A = {aij } be an n m matrix of rank r. We wish to nd a

second matrix B = {bij } of the same size, n m, but of rank t,

where t r, such that the least squares criterion, i,j (aij bij )2, is
as small as possible overall all matrices of rank t.

47
 The solution is to rst nd the singular value decomposition of A
as UDV , where U is n r and has orthonormal columns, V is
m r and has orthonormal columns, and D is r r, diagonal, with
positive values d1 d2 dr > 0 along the main diagonal.
Then, B is dened as UDV, where we take the rst t columns of
U and V to obtain U and V, respectively, and the rst t values,
d1 dt, to form a diagonal matrix D.
The approximation of A by a rank t matrix B, has been one mech-
anism for representing the row and column objects dening A in a
low-dimensional space of dimension t through what can be generi-
cally labeled as a biplot (the prex bi refers to the representation
of both the row and column objects together in the same space).
Explicitly, the approximation of A and B can be written as

B = UDV = UDD(1)V = PQ ,
where is some chosen number between 0 and 1, P = UD and

is n t, Q = (D(1)V ) and is m t.
The entries in P and Q dene coordinates for the row and column
objects in a t-dimensional space that, irrespective of the value of
chosen, have the following characteristic:
If a vector is drawn from the origin through the ith row point and
the m column points are projected onto this vector, the collection of
such projections is proportional to the ith row of the approximating
matrix B. The same is true for projections of row points onto vectors
from the origin through each of the column points.

48
0.4.6 The Procrustes Problem

Procrustes (the subduer), son of Poseidon, kept an inn beneting

from what he claimed to be a wonderful all-tting bed. He lopped o
excessive limbage from tall guests and either attened short guests by
hammering or stretched them by racking. The victim tted the bed
perfectly but, regrettably, died. To exclude the embarrassment of an
initially exact-tting guest, variants of the legend allow Procrustes
two, dierent-sized beds. Ultimately, in a crackdown on robbers
and monsters, the young Theseus tted Procrustes to his own bed.
(Gower and Dijksterhuis, 2004)
Suppose we have two matrices, X1 and X2, each considered (for
convenience) to be of the same size, n p. If you wish, X1 and
X2 can be interpreted as two separate p-dimensional coordinate sets
for the same set of n objects. Our task is to match these two con-
gurations optimally, with the criterion being least-squares: nd a
transformation matrix, Tpp , such that  X1T X2  is minimized,
where   denotes the sum-of-squares of the incorporated matrix,

i.e., if A = {auv }, then  A  = trace(AA) = u,v a2uv . For conve-
nience, assume both X1 and X2 have been normalized so  X1  =
 X2  = 1, and the columns of X1 and X2 have sums of zero.
Two results are central:
(a) When T is unrestricted, we have the multivariate multiple
regression solution
T = (X1X1)1X1X2 ;
(b) When T is orthogonal, we have the Schonemann solution done
for his thesis in the Quantitative Division at Illinois in 1965 (pub-
lished in Psychometrika in 1966):
49
 for the SVD of X2X1 = USV, we let T = VU.

0.4.7 Matrix Rank Reduction

Lagranges Theorem (as inappropriately named by C. R. Rao, be-

cause it should really be attributed to Guttman) can be stated as
follows:
Let G be a nonnegative-denite (i.e., a symmetric positive semi-
denite) matrix of order n n and of rank r > 0. Let B be of order
n s and such that BGB is non-singular. Then the residual matrix
G1 = G GB(BGB)1BG (1)
is of rank r s and is nonnegative denite.
Intuitively, this theorem allows you to take out factors from a
covariance (or correlation) matrix.

0.4.8 Torgerson Metric Multidimensional Scaling

Let A be a symmetric matrix of order n n. Suppose we want

to nd a matrix B of rank 1 (of order n n) in such a way that
the sum of the squared discrepancies between the elements of A and
 
the corresponding elements of B (i.e., nj=1 ni=1(aij bij )2) is at a
minimum. It can be shown that the solution is B = kk (so all
columns in B are multiples of k), where is the largest eigenvalue of
A and k is the corresponding normalized eigenvector. This theorem
can be generalized. Suppose we take the rst r largest eigenvalues
and the corresponding normalized eigenvectors. The eigenvectors are
collected in an nr matrix K = {k1, . . . , kr } and the eigenvalues in
a diagonal matrix . Then KK is an n n matrix of rank r and
50
is a least-squares solution for the approximation of A by a matrix of
rank r. It is assumed, here, that the eigenvalues are all positive. If
A is of rank r by itself and we take the r eigenvectors for which the
eigenvalues are dierent from zero collected in a matrix K of order
n r, then A = KK . Note that A could also be represented by
A = LL, where L = K1/2 (we factor the matrix), or as a sum of
r n n matrices A = 1k1k1 + + r krkr.
Metric Multidimensional Scaling Torgersons Model (Gowers
Principal Coordinate Analysis)
Suppose I have a set of n points that can be perfectly repre-
sented spatially in r dimensional space. 
The ith point has coordi-
r
k=1 (xik xjk ) represents the
nates (xi1, xi2, . . . , xir ). If dij = 2

Euclidean distance between points i and j, then

r

dij = xik xjk , where
k=1

1
dij = (d2ij Ai Bj + C); (2)
2
n

Ai = (1/n) d2ij ;
j=1

n

Bj = (1/n) d2ij ;
i=1

n n
2  
C = (1/n ) d2ij .
i=1 j=1

Note that {dij }nn = XX, where X is of order n r and the

entry in the ith row and k th column is xik .

51
 So, the Question: If I give you D = {dij }nn, nd me a set of
coordinates to do it. The Solution: Find D = {dij }, and take its
Spectral Decomposition. This is exact here.
To use this result to obtain a spatial representation for a set of
n objects given any distance-like measure, pij , between objects i
and j, we proceed as follows:
(a) Assume (i.e., pretend) the Euclidean model holds for pij .
(b) Dene pij from pij using (1).
(c) Obtain a spatial representation for pij using a suitable value
for r, the number of dimensions (at most, r can be no larger than
the number of positive eigenvalues for {pij }nn ):

{pij } XX
(d) Plot the n points in r dimensional space.

0.4.9 A Guttman Multidimensional Scaling Result

I. If B is a symmetric matrix of order n, having all its elements non-

negative, the following quadratic form dened by the matrix A must
be positive semi-denite:
 
bij (xi xj )2 = xiaij xj ,
i,j i,j

where

n
k=1;k=i bik (i = j)

aij =
bij (i = j)
If all elements of B are positive, then A is of rank n 1, and has
one smallest eigenvalue equal to zero with an associated eigenvector
52
having all constant elements. Because all (other) eigenvectors must
be orthogonal to the constant eigenvector, the entries in these other
eigenvectors must sum to zero.
This Guttman result can be used for a method of multidimensional
scaling (mds), and is one that seems to get reinvented periodically
in the literature. Generally, this method has been used to provide
rational starting points in iteratively-dened nonmetric mds.

0.4.10 A Few General MATLAB Routines to Know About

For Eigenvector/Eigenvalue Decompositions:

[V, D] = eig(A), where A = VDV, for A square; V is or-
thogonal and contains eigenvectors (as columns); D is diagonal and
contains the eigenvalues (ordered from smallest to largest ).
For Singular Value Decompositions:
[U, S, V] = svd(B), where B = USV; the columns of U and
the rows of V are orthonormal; S is diagonal and contains the non-
negative singular values (ordered from largest to smallest).
The help comments for the Procrustes routine in the Statistics
Toolbox are given verbatim below. Note the very general transfor-
mation provided in the form of a MATLAB Structure that involves
optimal rotation, translation, and scaling.
>> help procrustes
PROCRUSTES Procrustes Analysis
D = PROCRUSTES(X, Y) determines a linear transformation (translation,
reflection, orthogonal rotation, and scaling) of the points in the
matrix Y to best conform them to the points in the matrix X. The
"goodness-of-fit" criterion is the sum of squared errors. PROCRUSTES
returns the minimized value of this dissimilarity measure in D. D is
standardized by a measure of the scale of X, given by

53
 sum(sum((X - repmat(mean(X,1), size(X,1), 1)).^2, 1))

i.e., the sum of squared elements of a centered version of X. However,

if X comprises repetitions of the same point, the sum of squared errors
is not standardized.

X and Y are assumed to have the same number of points (rows), and
PROCRUSTES matches the ith point in Y to the ith point in X. Points
in Y can have smaller dimension (number of columns) than those in X.
In this case, PROCRUSTES adds columns of zeros to Y as necessary.

[D, Z] = PROCRUSTES(X, Y) also returns the transformed Y values.

[D, Z, TRANSFORM] = PROCRUSTES(X, Y) also returns the transformation

that maps Y to Z. TRANSFORM is a structure with fields:
c: the translation component
T: the orthogonal rotation and reflection component
b: the scale component
That is, Z = TRANSFORM.b * Y * TRANSFORM.T + TRANSFORM.c.

Examples:

% Create some random points in two dimensions

n = 10;
X = normrnd(0, 1, [n 2]);

% Those same points, rotated, scaled, translated, plus some noise

S = [0.5 -sqrt(3)/2; sqrt(3)/2 0.5]; % rotate 60 degrees
Y = normrnd(0.5*X*S + 2, 0.05, n, 2);

% Conform Y to X, plot original X and Y, and transformed Y

[d, Z, tr] = procrustes(X,Y);
plot(X(:,1),X(:,2),rx, Y(:,1),Y(:,2),b., Z(:,1),Z(:,2),bx);

% Compute a procrustes solution that does not include scaling:

trUnscaled.T = tr.T;
trUnscaled.b = 1;
trUnscaled.c = mean(X) - mean(Y) * trUnscaled.T;
ZUnscaled = Y * trUnscaled.T + repmat(trUnscaled.c,n,1);
dUnscaled = sum((ZUnscaled(:)-X(:)).^2) ...
/ sum(sum((X - repmat(mean(X,1),n,1)).^2, 1));

54