mca_cols = df.select_dtypes(['category']).columns
# instantiate MCA class
mca = prince.MCA(n_components = 3)
# get principal components
mca = mca.fit(df[mca_cols])
mca.explained_inertia_
mca.total_inertia_
I am attempting to do MCA with intention to feed the components into a clustering algorithm. However, when I look at the total_inertia_ I get a value of appx 70 and the explained inertia values for the 3 components are as follows:
[0.004080415064925553, 0.0035643891792985346, 0.003260362150059927]
The total inertia seems higher than what I've seen in examples for this library and the explained inertia values seem much lower. Can someone please explain what the reason might be and how I should interpret the total inertia value?
I was expecting total inertia to be around 1 and expected inertia values to add up to 1.
And another question, how do I find the ideal number of components for MCA?
