Determining the number of principal components using a scree plot
As we only need to retain the principal components that account for most of the variance of the original features, we can either use the Kaiser method, a scree plot, or the percentage of variation explained as the selection criteria. The main purpose of a scree plot is to graph the component analysis results as a scree plot and find where the obvious change in slope (elbow) occurs. In this recipe, we will demonstrate how to determine the number of principal components using a scree plot.
Getting ready
Ensure you have completed the previous recipe by generating a principal component object and saving it in variable eco.pca.
How to do it…
Perform the following steps to determine the number of principal components with a scree plot:
- First, generate a bar plot by using
screeplot:> screeplot(swiss.pca, type="barplot")
Figure 16: The scree plot in bar plot form
- You can also generate a line plot by using
screeplot...
