How many components are relevant?
The results that you stored in pcaNews contain as many components as variables. But your goal is a dimension reduction. It's time to find out how many components you should extract. Use several approaches to make your decision.
The results of the pca pcaNews are still loaded in your workspace. All necessary packages are loaded.
This exercise is part of the course
Machine Learning for Marketing Analytics in R
Exercise instructions
- Create a screeplot. How many components does it suggest?
- How many components would you need to meet the criterion of 70% variance explained (take a look at the
summary())? - How many components would you extract according to the Kaiser-Guttmann criterion?
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Screeplot:
___(___, type = ___)
# Cumulative explained variance:
___(___)
# Kaiser-Guttmann (number of components with eigenvalue larger than 1):
sum(pcaNews$___ > ___)