File EXAMPLE_C.dat

This example can be used to show the power of using 
mapping weights.  First find the Euclidian/Stress solution 
and then apply strong mapping weights.  While the two 
"pictures" are significantly different they are both "right."


This data set also provides a good example of how to 
deal with local minima.  It can be rather tedious to find
the global minimia (answer: Normalized Stress = 0.006251 for
ratio MDS, two dimensions, Euclidean distances)
because the problem has so many nearly idential 
local minima.  

TITLE=DEPARTMENT STORE PERCEPTION FACTORS
NObjects=10

DissimilarityList
Handy,  0
Speed,  .21, 0
Clean,  .59, .68, 0
Organ,  .74, .79, .2, 0
Junky,  .88, .8, .24, .25, 0
Times,  .11, .1, .66, .7, .89, 0
Close,  .13, .17, .6, .72, .77, .22, 0
Atmos,  .63, .69, .18, .22, .26, .7,  .71, 0
Decor,  .68, .65, .22, .19, .23, .61, .74, .18, 0
Large,  .82, .77, .28, .2, .17, .84, .83, .22, .23, 0 


Churchill used this data to find the relationships between various
characteristics of department stores.  The research question 
was posed in an unusual way.  Usually one would look for 
groupings of stores (i.e. object=store) and use the customer's
perceived characteristics as attributes.  Then one would 
calculate proximities between stores and make an MDS analysis.
However, Churchill choose to focus on the characteristics instead.  
So, he defined, Handy, Speed, ... as objects and calculated the 
correlation between them using a set of department
stores.  

Using factor analysis, Churchill concluded that there are two
main factors, Convenience and Atmosphere.  You will 
find the same thing using MDS, but the interfactor relationships 
will be more clearly shown than when factor analysis is used.






















