Team B Day 3- Mujtaba Mir

How to make a Dendogram!

We learnt a new technique of data analysis using Hierarchical Clustering.

The critical factor while considering the Clustering is the Dendogram. Hierarchical Clustering should be used when the number of variables under consideration is less than 50 in number.  Today we saw that we could group the variables to form different groups and create a story out of it. We clustered the functions together based on some form of commonality.

We used the Hierarchical Function. To reach the Hierarchical Cluster in SPSS go to Analyze

Then select the Statistics option and check Agglomeration:

Click Continue and Proceed to Plot option and make sure to check Dendogram

We used Jaccard Method to do the analysis today. Make sure the distance is Euclidean Distance and the cluster method is Between-Linkage. We also need to change the absent value to 2 as the data requires it to be so.

The Dendogram:

HIERARCHICAL CLUSTER   ANALYSIS Example

Statistical Method of determining how many clusters exist:

We need to draw a vertical line on the Dendogram based on how large the distance is when a new object joins to form a cluster. The line should be drawn where the distance is larger.

 We also learnt about the Jaccard Index:

JI =                Yes Matches_______

          (Total Matches- No Matches)

The Dendogram shows us which clusters are formed first and which cluster are closely linked together. In today’s class we came to know that how from a Dendogram we can bundle different VAS from the point of view of mobile service providers.