SPSS Tutorial

AEB 37 / AE 802
Marketing Research Methods
Week 7
 Cluster analysis
Lecture / Tutorial outline
Cluster analysis
Example of cluster analysis
Work on the assignment
 Cluster Analysis
It is a class of techniques used to
classify cases into groups that are
relatively homogeneous within
themselves and heterogeneous
between each other, on the basis of
a defined set of variables. These
groups are called clusters.
 Cluster Analysis and
marketing research
Market segmentation. E.g. clustering of
consumers according to their attribute
preferences
Understanding buyers behaviours.
Consumers with similar
behaviours/characteristics are clustered
Identifying new product opportunities.
Clusters of similar brands/products can help
identifying competitors / market opportunities
Reducing data. E.g. in preference mapping
 Steps to conduct a
Cluster Analysis
1. Select a distance measure
2. Select a clustering algorithm
3. Determine the number of clusters
4. Validate the analysis
 3

2
1

1
REGR factor score 1 for analysis

-1

-2

-3

-4
-3 -2 -1 0 1 2 3 4

REGR factor score 2 for analysis 1

 Defining distance: the
Euclidean distance
n
2
Dij x
k 1
ki xkj

Dij distance between cases i and j

xki value of variable Xk for case j
Problems:
Different measures = different weights
Correlation between variables (double
counting)
Solution: Principal component analysis
 Clustering procedures
Hierarchical procedures
Agglomerative (start from n clusters,
to get to 1 cluster)
Divisive (start from 1 cluster, to get to
n cluster)
Non hierarchical procedures
K-means clustering
Agglomerative clustering
 Agglomerative
clustering
Linkage methods
Single linkage (minimum distance)
Complete linkage (maximum distance)
Average linkage
Wards method
1. Compute sum of squared distances within clusters
2. Aggregate clusters with the minimum increase in the
overall sum of squares
Centroid method
The distance between two clusters is defined as the
difference between the centroids (cluster averages)
 K-means clustering
1. The number k of cluster is fixed
2. An initial set of k seeds (aggregation centres) is
provided
First k elements
Other seeds
3. Given a certain treshold, all units are assigned to
the nearest cluster seed
4. New seeds are computed
5. Go back to step 3 until no reclassification is
necessary
Units can be reassigned in successive steps
(optimising partioning)
 Hierarchical vs Non
hierarchical methods
Hierarchical Non hierarchical
clustering
clustering
No decision about the
number of clusters Faster, more reliable
Problems when data Need to specify the
contain a high level of number of clusters
error (arbitrary)
Can be very slow Need to set the initial
Initial decision are seeds (arbitrary)
more influential (one-
step only)
 Suggested approach
1. First perform a hierarchical
method to define the number of
clusters
2. Then use the k-means procedure
to actually form the clusters
 Defining the number of
clusters: elbow rule (1)
Agglomeration Schedule
n
Stage Cluster First
Stage Number of clusters Cluster Combined Appears
0 12 StageCluster 1 Cluster 2CoefficientsCluster 1 Cluster 2Next Stage
1 11 1 4 7 .015 0 0 4
2 10 2 6 10 .708 0 0 5
3 9 3 8 9 .974 0 0 4
4 8 4 4 8 1.042 1 3 6
5 7 5 1 6 1.100 0 2 7
6 6 6 4 5 3.680 4 0 7
7 5 7 1 4 3.492 5 6 8
8 4 8 1 11 6.744 7 0 9
9 3 9 1 2 8.276 8 0 10
10 2 10 1 12 8.787 9 0 11
11 1 11 1 3 11.403 10 0 0
 Elbow rule (2): the
scree diagram
12

10

8
Distance

0
11 10 9 8 7 6 5 4 3 2 1
Number of clusters
 Validating the
analysis
Impact of initial seeds / order of
cases
Impact of the selected method
Consider the relevance of the
chosen set of variables
SPSS Example
 1.5 MATTHEW
JULIA

1.0 LUCY
JENNIFER
.5 NICOLE

0.0

JOHN
-.5 PAMELA
THOMAS ARTHUR

-1.0
Component2

-1.5 FRED

-2.0
-1.5 -1.0 -.5 0.0 .5 1.0 1.5 2.0

Component1
 Agglomeration Schedule

Stage Cluster First

Cluster Combined Appears
Stage Cluster 1 Cluster 2 Coefficients Cluster 1 Cluster 2 Next Stage
1 3 6 .026 0 0 8
2 2 5 .078 0 0 7
3 4 9 .224 0 0 5
4 1 7 .409 0 0 6
5 4 10 .849 3 0 8
6 1 8 1.456 4 0 7
7 1 2 4.503 6 2 9
8 3 4 9.878 1 5 9
9 1 3 18.000 7 8 0

Number of clusters: 10 6 = 4
 1.5 MATTHEW
JULIA

1.0 LUCY
JENNIFER
.5 NICOLE

0.0

JOHN
-.5 PAMELA
THOMAS ARTHUR
Cluster Number of Ca

-1.0 4
Component2

3
-1.5 FRED
2

-2.0 1
-1.5 -1.0 -.5 0.0 .5 1.0 1.5 2.0

Component1
 Open the dataset
supermarkets.sav
From your N: directory (if you saved it
there last time
Or download it from:
http://www.rdg.ac.uk/~aes02mm/
supermarket.sav
Open it in SPSS
The supermarkets.sav
dataset
 Run Principal
Components Analysis
and save scores
Select the variables to perform the
analysis
Set the rule to extract principal
components
Give instruction to save the
principal components as new
variables
 Cluster analysis:
basic steps
Apply Wards methods on the
principal components score
Check the agglomeration schedule
Decide the number of clusters
Apply the k-means method
Analyse / Classify
 Select the component
scores

Select from here Untick this

Select Wards algorithm

Select
method here

Click here
first
Output: Agglomeration
schedule
 Number of clusters
Identify the step where the distance coefficients makes a bigger
jump
 The scree diagram
(Excel needed)
Distance

800

700

600

500

400

300

200

100

0
118

120

122

124

126

128

130

132

134

136

138

140

142

144

146

148
Step
 Number of clusters
Number of cases 150
Step of elbow 144
__________________________________
Number of clusters 6
 Now repeat the
analysis
Choose the k-means technique
Set 6 as the number of clusters
Save cluster number for each case
Run the analysis
K-means
K-means dialog box

Specify
number of
clusters
Save cluster membership

Click here
first Thick here
Final output
Cluster membership
 Component meaning
(tutorial week 5)
4. Organic radio
Component Matrixa
listener
1. Old Rich Big
Component
Spender 3. Vegetarian TV
1 2 3 4 5
Monthly amount spent .810 lover
-.294 -4.26E-02 .183 .173
Meat expenditure
2. Family
.480
shopper
-.152 .347 .334 -5.95E-02
Fish expenditure .525 -.206 -.475 -4.35E-02 .140
Vegetables expenditure .192 -.345 -.127 .383 5. Vegetarian
.199 TV and
-.207web hater
% spent in own-brand
.646 -.281 -.134 -.239
product
Own a car .536 .619 -.102 -.172 6.008E-02
% spent in organic food .492 -.186 .190 .460 .342
Vegetarian 1.784E-02 -9.24E-02 .647 -.287 .507
Household Size .649 .612 .135 -6.12E-02 -3.29E-03
Number of kids .369 .663 .247 .184 1.694E-02
Weekly TV watching
.124 -9.53E-02 .462 .232 -.529
(hours)
Weekly Radio listening
2.989E-02 .406 -.349 .559 -8.14E-02
(hours)
Surf the web .443 -.271 .182 -5.61E-02 -.465
Yearly household income .908 -4.75E-02 -7.46E-02 -.197 -3.26E-02
Age of respondent .891 -5.64E-02 -6.73E-02 -.228 6.942E-04
Extraction Method: Principal Component Analysis.
a. 5 components extracted.
 Final Cluster Centers

Cluster
1 2 3 4 5 6
REGR factor score
-1.34392 .21758 .13646 .77126 .40776 .72711
1 for analysis 1
REGR factor score
.38724 -.57755 -1.12759 .84536 .57109 -.58943
2 for analysis 1
REGR factor score
-.22215 -.09743 1.41343 .17812 1.05295 -1.39335
3 for analysis 1
REGR factor score
.15052 -.28837 -.30786 1.09055 -1.34106 .04972
4 for analysis 1
REGR factor score
.04886 -.93375 1.23631 -.11108 .31902 .87815
5 for analysis 1
 Cluster interpretation
through mean component values
Cluster 1 is very far from profile 1 (-1.34) and
more similar to profile 2 (0.38)
Cluster 2 is very far from profile 5 (-0.93) and
not particularly similar to any profile
Cluster 3 is extremely similar to profiles 3 and 5
and very far from profile 2
Cluster 4 is similar to profiles 2 and 4
Cluster 5 is very similar to profile 3 and very far
from profile 4
Cluster 6 is very similar to profile 5 and very far
from profile 3
 Which cluster to
target?
Objective: target the organic
consumer
Which is the cluster that looks more
organic?
Compute the descriptive statistics
on the original variables for that
cluster
Representation of factors 1
and 4
(and cluster membership)
3

2
1
REGR factor score 4 for analysis

Cluster Number of Ca
0
6

5
-1
4

3
-2
2

-3 1
-3 -2 -1 0 1 2

REGR factor score 1 for analysis 1