Está en la página 1de 7

Principle Component Analysis (PCA)

Objective:
The objective of the principal component analysis is to

Reduce the number of variables from 31 to fewer composite components with minimum loss of information.
Reduce the multi-collinearity which is the pre-requisite for cluster analysis.

Design:
The sample size of 106 provides an adequate basis for the calculation of the correlations between the variables. Also the
sample size ratio of 3.5:1 of observations to variables falls within acceptable limits.
Assumptions:

All observed variables satisfy multicollinearity, homoscedasticity and normality.


The correlation matrix has 91 out of 465 correlations (20%) significant at 0.05 level, which provides adequate
basis for proceeding to an empirical examination of adequacy for factor analysis on both an overall basis and for
each variable.
The overall significance of the correlation matrix with the Bartlett test and factorability of the overall set of
variables and individuals variables using the measure of sampling adequacy (MSA).

Limitations:

SPSS v21 doesnt generate the KMO statistics. Hence it is assumed that it meets the minimum MSA requirement
of 0.6 and above to proceed with PCA analysis.
Factor analysis will always derive factors.
Variables that arent included and may have possible effects in their absence has on the resulting factors.

Communalities
Initial
Lowincarbs
Tastesgreat
Dietplan
Energybooster
Antioxidant
Digestionaid
Vitaminsupplement
Package
Sweetness
Thickness
Calories
Vitamins
Carbonated
Nopreservatives
Price
BrandImage
Advertisements
Availability
Usage_everyday
Usage_Monthly
Usage_weekly
Usage_Never
PS_2L
PS_1L
PS_05L
PS_2L
PS_01L
PL_LT20
PL_2025
PL_2630
PLGT30

Extraction

1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000

.253
.763
.383
.313
.359
.284
.417
.533
.657
.497
.565
.657
.534
.415
.537
.566
.513
.600
.457
.807
.519
.603
.339
.666
.455
.692
.220
.717
.780
.633
.407

Extraction Method: Principal Component


Analysis.

Communality (h) is the sum of the squared component loadings and represents the amount of variance in that variable
accounted for by all the components. This is the proportion of each variable's variance that can be explained by the
principal components.
All seven extracted components account for 25.3% of the variance in variable y1 (h = .253).
The values in this column indicate the proportion of each variable's variance that can be explained by the principal
components. Variables with high values are well represented in the common factor space, while variables with low
values are not well represented. Here, there are no particularly low values.

Total Variance Explained


Component

Initial Eigenvalues
Total

1
2
3
4
5
6
7

% of
Variance

Extraction Sums of Squared


Loadings

Cumulative
%

4.886
2.522
2.115
1.813
1.737
1.567
1.503
1.419

15.760
8.135
6.823
5.849
5.602
5.054
4.850
4.578

15.760
23.894
30.717
36.566
42.168
47.222
52.072
56.650

1.308

4.219

60.869

10

1.278

4.124

64.993

11

1.137

3.667

68.659

12

1.069

3.449

72.108

13

1.047

3.377

75.486

14

.930

3.002

78.487

15

.874

2.820

81.307

16

.802

2.588

83.895

17

.791

2.553

86.448

18

.695

2.242

88.690

19

.608

1.962

90.651

20

.579

1.869

92.520

21

.513

1.656

94.177

22

.409

1.320

95.497

23

.353

1.138

96.634

24

.334

1.076

97.710

25

.263

.850

98.560

26

.241

.778

99.338

27

.205

.662

100.000

1.017E- 1.053E-013
013
1.002E- 1.007E-013
013
-1.004E-1.014E013
013
-1.007E-1.022E013
013

100.000

28
29
30
31

Total
4.886
2.522
2.115
1.813
1.737
1.567
1.503

% of
Variance
15.760
8.135
6.823
5.849
5.602
5.054
4.850

Cumulative
%
15.760
23.894
30.717
36.566
42.168
47.222
52.072

Rotation Sums of Squared Loadings


Total
4.837
2.221
1.991
1.829
1.799
1.788
1.677

% of
Variance
15.602
7.165
6.423
5.900
5.805
5.767
5.410

Cumulative
%
15.602
22.767
29.190
35.090
40.895
46.662
52.072

100.000
100.000
100.000

Extraction Method: Principal Component Analysis.

There are as many components extracted during a principal components analysis as there are variables that are put into
it. There are 31 variables and hence 31 components. Eigenvalues are the variances of the principal components.
Because the principal components analysis on the correlation matrix, the variables are standardized, which means that
the each variable has a variance of 1, and the total variance is equal to the number of variables used in the analysis, in
this case, 31.

The column total contains the eigenvalues. The first component will always account for the most variance (and hence
have the highest eigenvalue), and the next component will account for as much of the left over variance as it can, and so
on. Hence, each successive component will account for less and less variance.
The cumulative % contains the cumulative percentage of variance accounted for by the current and all preceding
principal components. The seventh row shows a value of 52.072. This means that the first seven components together
account for 52.072% of the total variance. All variance is considered to be true and common variance and the variables
are assumed to be measured without error, so there is no error variance.
Total 13 number of principal components whose eigenvalues are 1.0 or greater are extracted. But the difference in
variance decreases from the seventh component onwards. Due to non-linear decrease, we stop with the seventh
component.

The scree plot graphs the eigenvalue against the component number. From the seventh component on, the line is
almost flat with marginal increase, meaning the each successive component is accounting for smaller and smaller
amounts of the total variance. Components with an eigenvalue of less than 1.3 account for less variance than the
original variable and so are of little use. The point of principal components analysis is to redistribute the variance in the
correlation matrix using the method of eigenvalue decomposition to redistribute the variance to first components
extracted.

Rotated Component Matrix

Component
1

Availability

.741

Sweetness

.717

Thickness

.697

Package

.681

Price

.658

Vitamins

.653

BrandImage

.651

Calories

.577

Advertisements

.564

Nopreservatives

.511

Sweetness
& Thickness

PLGT30
PL_2025

-.807

PL_LT20

.794

Price

Usage_weekly
PS_01L
Usage_Monthly

-.819

PS_2L

-.520

Usage

Usage_Never
Carbonated
Tastesgreat
Energybooster

-.846
.536

-.729

PS_2L
PS_1L

Taste

Package size

.519

Lowincarbs
.706

PL_2630
Usage_everyday

Antioxidant

.510

Antioxidant
Vitaminsupplement
Dietplan
PS_05L
Digestionaid

Diet &
Digestion aid

-.577
.548
.515

Extraction Method: Principal Component Analysis.


a
Rotation Method: Varimax with Kaiser Normalization.
a. Rotation converged in 41 iterations.

Rotation maximizies the loading of each variable on one of the extracted variables while minimizing the loading on all
other factors. Varimax is the orthogonal rotation technique employed in the analysis.

Component Transformation Matrix


Component

.992

.016

-.021

-.045

.063

-.056

-.076

.013

-.810

.337

.407

.176

-.010

-.183

.026

-.181

-.757

.308

.296

.283

.362

-.047

.140

.155

-.272

.646

.582

-.353

.096

.162

.219

.403

-.511

.700

.059

-.022

.515

.136

.701

.354

-.295

-.112

.053

.019

.471

-.098

.274

.028

.830

Extraction Method: Principal Component Analysis.


Rotation Method: Varimax with Kaiser Normalization.

Component Transformation Matrix displays the correlations among the components prior to and after rotation. The
diagonal elements have fairly large correlation. Because we used an orthogonal rotation, this should be a diagonal
matrix, meaning that the same number should appear in all places along the diagonal. In actuality the factors are
uncorrelated; however, because factor scores are estimated there may be slight correlations among the factor scores.
The other elements have fairly low correlation.
Correlations
Correlations
significant at 0.05 level
Lowincarbs

Pearson Correlation
Sig. (2-tailed)

Tastesgreat

Pearson Correlation
Sig. (2-tailed)

Dietplan

Pearson Correlation
Sig. (2-tailed)

Carbonated

Pearson Correlation
Sig. (2-tailed)

Vitamins

10

Pearson Correlation
Sig. (2-tailed)

Calories

11

Pearson Correlation
Sig. (2-tailed)

Thickness

Pearson Correlation
Sig. (2-tailed)

Sweetness

Pearson Correlation
Sig. (2-tailed)

Package

Pearson Correlation
Sig. (2-tailed)

Vitaminsupplement

Pearson Correlation
Sig. (2-tailed)

Digestionaid

Pearson Correlation
Sig. (2-tailed)

Antioxidant

Pearson Correlation
Sig. (2-tailed)

Energybooster

Pearson Correlation
Sig. (2-tailed)

Nopreservatives

Pearson Correlation
Sig. (2-tailed)

Price

Sig. (2-tailed)
BrandImage

Pearson Correlation
Sig. (2-tailed)

PLGT30

Pearson Correlation
Sig. (2-tailed)

PL_2630

Pearson Correlation
Sig. (2-tailed)

PL_2025

Pearson Correlation
Sig. (2-tailed)

PL_LT20

Pearson Correlation
Sig. (2-tailed)

PS_01L

Pearson Correlation
Sig. (2-tailed)

PS_2L

Pearson Correlation
Sig. (2-tailed)

PS_05L

Pearson Correlation
Sig. (2-tailed)

PS_1L

Pearson Correlation
Sig. (2-tailed)

PS_2L

Pearson Correlation
Sig. (2-tailed)

Usage_Never

Pearson Correlation
Sig. (2-tailed)

Usage_weekly

Pearson Correlation
Sig. (2-tailed)

Usage_Monthly

Pearson Correlation
Sig. (2-tailed)

Usage_everyday

Pearson Correlation
Sig. (2-tailed)

Availability

Pearson Correlation
Sig. (2-tailed)

Advertisements

Pearson Correlation

Pearson Correlation
Sig. (2-tailed)

*. Correlation is significant at the 0.05 level (2-tailed).

0
91
20%

También podría gustarte