Agriculture Reference
In-Depth Information
this method lies on placing the objects/
individuals on a two-dimensional space taking
any two of the
options in the statistical packages. While dealing
with cluster analysis, the following points may be
noted:
1. Cluster analysis is a useful technique and can
be applied for univariate as well as multivari-
ate data.
2. Presence of outliers should be dealt with cau-
tion as most of the clustering techniques are
sensitive to the presence of outliers in the data.
3. Selection of appropriate similarity or dissimi-
larity measure should be taken with utmost
care.
4. Similarly, the linkage method to be adopted is
also not exclusive for a particular situation,
though average linkage is preferred over the
others.
5. As most of the clustering analyses are to be
performed using computer software and in
standard packages there are a number of
options in each stage, one should be very
careful in selecting the appropriate one.
Knowledge of the technique and methods is
essential to have meaningful conclusion from
cluster analysis.
characters. Next other characters
for each and every individual are presented with
the help of lines/rays or some other geometric
figures proportional to some suitably transformed
index/scales/scores (if required). From the place-
ment and the cumulative values (or scores) of the
individuals, one can have distinct idea about the
different groups. This method is also objected for
its subjective nature.
p
12.11.4 Dendrogram (Tree Diagram)
Graphical representation linking the objects or
variables based on the criteria discussed is
known as dendrogram or tree diagram.
Readers must note that clustering is a power-
ful multivariate technique helpful in grouping the
variables as well as individuals of any popula-
tion, but it requires extensive knowledge about
the theory of the method to be used, its useful-
ness, and appropriateness under specific situa-
tion. Researchers should be guided by the
conceptual framework about the subject con-
cerned and the requirement and should not be
tempted to be guided by the availability of the
Example 12.7.
The table in the next page gives
data on 19 yield component characters for 37
varieties. Using the data cluster the varieties.
