Principal Component Analysis
(PCA)
www.csce.uark.edu/~xintaowu/5073/PCA1.ppt
Data Reduction
summarization of data with many (p)
variables by a smaller set of (k) derived
(synthetic, composite) variables.
p
Data Reduction
Residual variation is information in A
that is not retained in X
balancing act between
clarity of representation, ease of
understanding
oversimplification: loss of important or
relevant information.
Principal Component Analysis
(PCA)
probably the most widely-used and wellknown of the standard multivariate
methods
invented by Pearson (1901) and Hotelling
(1933)
first applied in ecology by Goodall (1954)
under the name factor analysis (principal
factor analysis is a synonym of PCA).
Principal Component Analysis
(PCA)
takes a data matrix of n objects by p
variables, which may be correlated, and
summarizes it by uncorrelated axes
(principal components or principal axes)
that are linear combinations of the
original p variables
the first k components display as much as
possible of the variation among objects.
Geometric Rationale of PCA
objects are represented as a cloud of n
points in a multidimensional space with an
axis for each of the p variables
the centroid of the points is defined by
the mean of each variable
the variance of each variable is the
average squared deviation of its n values
around the mean of that variable.
1
2
X im X i
Vi
n 1 m 1
n
Geometric Rationale of PCA
degree to which the variables are linearly
correlated is represented by their
covariances.
1 n
X im X i X jm X j
Cij
n 1 m 1
Covariance of
variables i and j
Sum over all
n objects
Value of
Mean of
variable i variable i
in object m
Value of
variable j
in object m
Mean of
variable j
Geometric Rationale of PCA
objective of PCA is to rigidly rotate the
axes of this p-dimensional space to new
positions (principal axes) that have the
following properties:
ordered such that principal axis 1 has the
highest variance, axis 2 has the next highest
variance, .... , and axis p has the lowest
variance
covariance among each pair of the principal
axes is zero (the principal axes are
uncorrelated).
2D Example of PCA
variables X1 and X2 have positive covariance & each has a
similar variance.
X 2 4.91
X 1 8.35
V1 6.67
V2 6.24
C1, 2 3.42
Configuration is Centered
each variable is adjusted to a mean of
zero (by subtracting the mean from each value).
Principal Components are Computed
PC 1 has the highest possible variance (9.88)
PC 2 has a variance of 3.03
PC 1 and PC 2 have zero covariance.
The Dissimilarity Measure Used in PCA is
Euclidean Distance
PCA uses Euclidean Distance calculated
from the p variables as the measure of
dissimilarity among the n objects
PCA derives the best possible k dimensional
(k < p) representation of the Euclidean
distances among objects.
Generalization to p-dimensions
In practice nobody uses PCA with only 2
variables
The algebra for finding principal axes
readily generalizes to p variables
PC 1 is the direction of maximum variance
in the p-dimensional cloud of points
PC 2 is in the direction of the next
highest variance, subject to the
constraint that it has zero covariance
with PC 1.
Generalization to p-dimensions
PC 3 is in the direction of the next highest
variance, subject to the constraint that it
has zero covariance with both PC 1 and PC 2
and so on... up to PC p
each principal axis is a linear combination of the original
two variables
PCj = ai1Y1 + ai2Y2 + ainYn
aijs are the coefficients for factor i, multiplied by the
measured value for variable j
PC 1
PC 2
PC axes are a rigid rotation of the original variables
PC 1 is simultaneously the direction of maximum variance
and a least-squares line of best fit (squared distances
of points away from PC 1 are minimized).
PC 1
PC 2
Generalization to p-dimensions
if we take the first k principal components, they
define the k-dimensional hyperplane of best fit
to the point cloud
of the total variance of all p variables:
PCs 1 to k represent the maximum possible proportion of
that variance that can be displayed in k dimensions
i.e. the squared Euclidean distances among points
calculated from their coordinates on PCs 1 to k are the
best possible representation of their squared Euclidean
distances in the full p dimensions.
Covariance vs Correlation
using covariances among variables only
makes sense if they are measured in the
same units
even then, variables with high variances will
dominate the principal components
these problems are generally avoided by
standardizing each variable to unit variance
and zero mean.
X im
Xi
Standard deviation
of variable i
SD i
im
Mean
variable i
Covariance vs Correlation
covariances between the standardized
variables are correlations
after standardization, each variable has a
variance of 1.000
correlations can be also calculated from the
variances and covariances:
Correlation between
variables i and j
rij
Variance
of variable i
C ij
ViV j
Covariance of
variables i and j
Variance
of variable j
The Algebra of PCA
first step is to calculate the cross-products
matrix of variances and covariances (or
correlations) among every pair of the p
variables
square, symmetric matrix
diagonals are the variances, off-diagonals
are the covariances.
X1
X2
X1
X2
X1
6.6707
3.4170
X1
1.0000
0.5297
X2
3.4170
6.2384
X2
0.5297
1.0000
Variance-covariance Matrix
Correlation Matrix
The Algebra of PCA
in matrix notation, this is computed as
S XX
where X is the n x p data matrix, with each
variable centered (also standardized by SD if using
correlations).
X1
X2
X1
X2
X1
6.6707
3.4170
X1
1.0000
0.5297
X2
3.4170
6.2384
X2
0.5297
1.0000
Variance-covariance Matrix
Correlation Matrix
Manipulating Matrices
transposing: could change the columns to
rows or the rows to columns
X =
10 0 4
7 1 2
multiplying matrices
X =
10 7
0 1
4 2
must have the same number of columns in the
premultiplicand matrix as the number of rows in
the postmultiplicand matrix
The Algebra of PCA
sum of the diagonals of the variancecovariance matrix is called the trace
it represents the total variance in the data
it is the mean squared Euclidean distance
between each object and the centroid in pdimensional space.
X1
X2
X1
X2
X1
6.6707
3.4170
X1
1.0000
0.5297
X2
3.4170
6.2384
X2
0.5297
1.0000
Trace = 12.9091
Trace = 2.0000
The Algebra of PCA
finding the principal axes involves
eigenanalysis of the cross-products matrix
(S)
the eigenvalues (latent roots) of S are
solutions () to the characteristic equation
S I 0
The Algebra of PCA
the eigenvalues, 1, 2, ... p are the
variances of the coordinates on each
principal component axis
the sum of all p eigenvalues equals the trace
of S (the sum of the variances of the
original variables).
X1
X2
X1
6.6707
3.4170
X2
3.4170
6.2384
1 = 9.8783
2 = 3.0308
Note:
=12.9091
1
2
Trace = 12.9091
The Algebra of PCA
each eigenvector consists of p values which
represent the contribution of each
variable to the principal component axis
eigenvectors are uncorrelated (orthogonal)
their cross-products are zero.
Eigenvectors
u1
u2
X1
0.7291
-0.6844
X2
0.6844
0.7291
0.7291*(-0.6844) + 0.6844*0.7291 = 0
The Algebra of PCA
coordinates of each object i on the kth
principal axis, known as the scores on PC k,
are computed as
z ki u1k x1i u2 k x 2 i u pk x pi
where Z is the n x k matrix of PC scores, X
is the n x p centered data matrix and U is
the p x k matrix of eigenvectors.
The Algebra of PCA
variance of the scores on each PC axis is
equal to the corresponding eigenvalue for
that axis
the eigenvalue represents the variance
displayed (explained or extracted) by
the kth axis
the sum of the first k eigenvalues is the
variance explained by the k-dimensional
ordination.
1 = 9.8783
2 = 3.0308
Trace = 12.9091
PC 1 displays (explains)
9.8783/12.9091 = 76.5% of the total variance
The Algebra of PCA
The cross-products matrix computed among
the p principal axes has a simple form:
all off-diagonal values are zero (the principal
axes are uncorrelated)
the diagonal values are the eigenvalues.
PC1
PC2
PC1
9.8783
0.0000
PC2
0.0000
3.0308
Variance-covariance Matrix
of the PC axes
A more challenging example
data from research on habitat definition in
the endangered Baw Baw frog
16 environmental and structural variables
measured at each of 124 sites
correlation matrix used because variables
have different units
Philoria frosti
Eigenvalues
Axis
Eigenvalue
% of
Variance
Cumulative %
of Variance
5.855
36.60
36.60
3.420
21.38
57.97
1.122
7.01
64.98
1.116
6.97
71.95
0.982
6.14
78.09
0.725
4.53
82.62
0.563
3.52
86.14
0.529
3.31
89.45
0.476
2.98
92.42
10
0.375
2.35
94.77
Interpreting Eigenvectors
correlations
between variables
and the principal
axes are known as
loadings
each element of
the eigenvectors
represents the
contribution of a
given variable to a
component
0.3842
0.0659
-0.1177
pH
-0.1159
0.1696
-0.5578
Cond
-0.2729
-0.1200
0.3636
0.0538
-0.2800
0.2621
-0.0765
0.3855
-0.1462
maxERht
0.0248
0.4879
0.2426
avERht
0.0599
0.4568
0.2497
%ER
0.0789
0.4223
0.2278
%VEG
0.3305
-0.2087
-0.0276
%LIT
-0.3053
0.1226
0.1145
%LOG
-0.3144
0.0402
-0.1067
%W
-0.0886
-0.0654
-0.1171
0.1364
-0.1262
0.4761
DistSWH
-0.3787
0.0101
0.0042
DistSW
-0.3494
-0.1283
0.1166
Altitude
TempSurf
Relief
H1Moss
How many axes are needed?
does the (k+1)th principal axis represent
more variance than would be expected
by chance?
several tests and rules have been
proposed
a common rule of thumb when PCA is
based on correlations is that axes with
eigenvalues > 1 are worth interpreting
What are the assumptions of PCA?
assumes relationships among variables are
LINEAR
cloud of points in p-dimensional space has linear
dimensions that can be effectively summarized
by the principal axes
if the structure in the data is NONLINEAR
(the cloud of points twists and curves its
way through p-dimensional space), the
principal axes will not be an efficient and
informative summary of the data.
When should PCA be used?
In community ecology, PCA is useful for
summarizing variables whose relationships
are approximately linear or at least
monotonic
e.g. A PCA of many soil properties might be
used to extract a few components that
summarize main dimensions of soil variation
PCA is generally NOT useful for
ordinating community data
Why? Because relationships among
species are highly nonlinear.
The Horseshoe or Arch Effect
community trends along environmenal
gradients appear as horseshoes in PCA
ordinations
none of the PC axes effectively
summarizes the trend in species
composition along the gradient
SUs at opposite extremes of the gradient
appear relatively close together.
Ambiguity of Absence
Axis 2
Beta Diversity 2R - Covariance
Axis 1
The HorseshoeEffect
curvature of the gradient and the degree
of infolding of the extremes increase with
beta diversity
PCA ordinations are not useful summaries
of community data except when beta
diversity is very low
using correlation generally does better than
covariance
this is because standardization by species
improves the correlation between Euclidean
distance and environmental distance.
What if theres more than one underlying
ecological gradient?
The Horseshoe Effect
when two or more underlying gradients with
high beta diversity a horseshoe is usually
not detectable
the SUs fall on a curved hypersurface that
twists and turns through the p-dimensional
species space
interpretation problems are more severe
PCA should NOT be used with community
data (except maybe when beta diversity is very low).
Impact on Ordination History
by 1970 PCA was the ordination method of
choice for community data
simulation studies by Swan (1970) &
Austin & Noy-Meir (1971) demonstrated
the horseshoe effect and showed that the
linear assumption of PCA was not
compatible with the nonlinear structure of
community data
stimulated the quest for more appropriate
ordination methods.
