<< Click to Display Table of Contents >> Navigation:  2. Introduction to the PCA techniques > 2.3. Summary

 

To summarize, PCA is used to:
 

1.do dimensionality reduction.
 

2.create good « distances-definition ». We need a correct « distances-definition » so that the K-Mean algorithm works properly and delivers good segmentation models. Without a PCA, the classical « distances-definitions » base on Euclidean-distances are giving random weights to each concept, thereby producing a totally random and arbitrary segmentation.

 
 

As you can see, a correct methodology for a good segmentation analysis relies heavily on PCA analysis. Stardust is the only segmentation tool that is directly offering you, in a few mouse clicks, a complete « PCA analysis » in all the steps of the segmentation analysis.

 

 
We will now see how these two concepts are used inside StarDust to:
 

•Produce a segmentation model
 

•Describe each segments from a business perspective