Beruflich Dokumente
Kultur Dokumente
REFERENCES
Hair. Multivariate Data Analysis
L.J. Stalans (1998) Multi-dimensional
WHAT IS MDS
As a perceptual mapping (spatial map) of objects
WHAT IS MDS
Data : approximate the distances between pairs of the
WHAT IS MDS
Euclidean distance to
model dissimilarity,
position of point i on
dimension a:
Metric MDS: the
similarities had to be
quantitative
Non metric MDS: data
is ordinal
dij
(x
ia
x ja )
Distance
d12 a 2 b 2
Example
Data: consumers perceptions of six candy bars.
AC
1.0
.5
candy f
candy e
candy d
0.0
D im e n sio n 2
-.5
-1.0
-1.5
candy c
-2.0
-2.5
-2.0
Dimension 1
-1.5
-1.0
-.5
0.0
.5
1.0
1.5
Procedures
1. Array the objects along the straight line
2. Make a proximity values (amount of similarity or
n(n 1)
2
Stress Index
Stress
(d ij dij ) 2
( d ij d ) 2
Goodness of fit
poor
fair
good
excellent
perfect
APLICATION
MDS helps the researchers to identify key
dimensions underlying
respondentsevaluations of objects.
E.g:
identify key dimensions underlying
customer evaluations of products, services
or companies.
Comparison dimensions of objects
Aggregate Analysis
Fewer perceptual maps
Aggregations may take place either before
Chekitan
Market segmentation
Competitive market structure of certain
product
The perceptions of individual firms about
bank products, see: article of
Athanassopoulus.
judgements
Distance in a spatial plot is an analogy for
the perceived relatedness of items
Spatial plot called a configurations
Dimensions and co-ordinates
TYPES
Metric scaling, problem of data restrictions
Non metric scaling, most common, assumes data
Calculation of similarities
See file similarities
Suppose five individuals possess the following characteristics
Individual 1
Individual 2
Individual 3
Individual 4
Individual 5
height
68 in
73 in
67 9n
64 in
76 in
Weight
140 lb
185
165
120
210
Eye color
green
brown
blue
brown
brown
Hair color
blond
brown
blond
brown
brown
HandednessGender
right
female
right
male
right
male
right
female
left
male
Example of CA
Table 9-2, Table 9-3, Figure 9-11
Table 9-9, 9-10, Figure 9-17