A Gaussian mixture model represents data as a mixture of Gaussian component distributions. For clustering, it gives each point a fraction of membership in every existing cluster rather than assigning it to only one. section 2.11
Soft membership for an email
Suppose an email could belong to three groups, A, B, and C. The model might assign 60% membership to A, 25% to B, and 15% to C. section 2.11
Figure 2.5 A plot of the email dataset. The horizontal axis corresponds to the size of the email and the vertical axis to the number of recipients. We can see three well-defined clusters in this dataset.
A plot of email size against recipient count shows three visible groups, providing an intuitive setting for soft cluster assignments. section 2.11





