Cluster Analysis
What is cluster analysis?
Cluster analysis attempts to identify relatively homogeneous clusters of respondents in respect to selected variables , so that those within a cluster are as similar as possible to one another and people between clusters are as dissimilar as possible just.
This is exactly what marketers want an ideal market segment to be when attempting to divide the total heterogeneous market into more homogeneous markets for targeting.
Case Study
Suppose a wireless communications company wants to explore the issue of market segmentation using the data from the consumer survey. It believes that the market can be segmented on the basis of the eight attributes: importance responses. The eight attributes are range, mobility, sound, place, preciev, avgbill, telephone, install.
The questions in the questionnaire may look like this: (VI = Very important, VU=Very Unimportance)
Asian B2B Expert
China | Hong Kong | Japan | Korea | Taiwan | Singapore | India | Australia
What you give us?
The ratings data which may be stored in an SPSS or Excel file using the following format:
What outputs are produced to you?
(1)
Dendrogram and agglomeration schedule chart will be shown to help you understand how the number of clusters are determined. Dendrogram is a tree figure that indicates the groups of cases/objects formed at various similarity (distance) levels.
(2)
Suggest number of clusters and the profile of each
Suppose the optimal numbers of clusters is found to be three, their sizes and means ratings on the eight attributes are compared and summarized in the following table and chart.
(3)
Interpretation and Application:
Cluster 3 is the largest, with 83 respondents, and cluster 2 is the smallest, with 55 respondents.
In regard to attribute importance ratings, clusters 2 and 3 are the most similar, They have almost identical average importance ratings on six of the eight attributes. They do differ significantly in regard to price sensitivity (TELEPHON and INSTALL). Members of cluster 3 assign a much higher importance to the price of the required equipment and installation/activation charges.
Cluster 1 differs markedly from clusters 2 and 3. In particular, members of this cluster view RANGE, MOBILITY, PLACE, and PRECEIV to be significantly less important than do the members of the other two clusters
Members of clusters 2 and 3 are the best targets for the new wireless service, based on the importance they place on the RANGE, MOBILITY PRECEIV, and PLACE variables. In the early stages of product introduction, cluster 3 members represent the most attractive target because of their relative insensitivity to price. Cluster 2 members might be targeted later, assuming that prices fall over time, which is the projected scenario.
The company's initial target should be members of cluster 3, for the reasons just mentioned. Later, as prices fall, an appropriate strategy might be to expand the target to include the members of cluster 2. Member of cluster 1 should be targeted only after the company has achieved saturation penetration levels among members of clusters 2 and 3
Attribute
|
VI
|
VU
|
|||||||
range of coverage
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
motility
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
sound quality
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
ability to place calls when away from home
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
ability to place and receive calls when away from home
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
average monthly telephone bill
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
price of telephone equipment
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
installation charge
|
9
|
8
|
7
|
6
|
5
|
4
|
3
|
2
|
1
|
Respondent ID
|
range
|
mobility
|
sound
|
place
|
precieve
|
avgbill
|
telephone
|
install
|
1
|
2
|
7
|
5
|
9
|
8
|
3
|
7
|
3
|
2
|
5
|
9
|
9
|
5
|
9
|
4
|
4
|
3
|
3
|
5
|
9
|
9
|
4
|
8
|
3
|
3
|
3
|
4
|
3
|
6
|
6
|
8
|
8
|
3
|
6
|
4
|
