For each question, tick the correct answer or answers.

Q1) Between-class inertia.
The sum of the losses of between-class inertia is equal to the total inertia
When the variables are standardized, the sum of the losses of between-class inertia equals the number of variables

Q2) In a diagram showing between-class inertia,
the first bar (the largest) shows the loss of between-class inertia when moving from 2 clusters to 1
the first bar (the largest) shows the increase in between-class inertia when moving from 2 clusters to 1
the first bar (the largest) gives the increase in within-class inertia when moving from 2 clusters to 1
losses in between-class inertia decrease in size as we split the data into more and more clusters

Q3) Partition quality
A partition into few clusters risks ending up with heterogeneous clusters
A partition into many clusters risks ending up with heterogeneous clusters
A good partition means ending up with homogeneous and interpretable classes

Q4) The choice of the number of clusters
may become clear by looking at the hierarchical tree
may be made clear by looking at the diagram showing losses in between-class inertia
must lead to homogeneous and interpretable classes

Q5) Between-class inertia.
The between-class inertia of a clustering with only 1 cluster is 0
The between-class inertia of a clustering into 2 clusters is always less than the inertia of the 1st PCA axis
The between-class inertia of a clustering into 3 clusters is always less than the inertia of the 1st PCA plane

Score =
Correct answers: