Highly Correlated Variables
Generally, two highly correlated variables likely contribute to the prediction ability of the model, which makes one redundant. For example, if we have a dataset with age, height, and BMI as variables, we know that BMI is a function of age and height and it will always be highly correlated with the other two. If it's not, then something is wrong with the BMI calculation. In such cases, one might decide to remove the other two. However, it is always not this straight. In certain cases, a pair of variables might be highly correlated, but it is not easy to interpret why that is the case. In such cases, one can randomly drop one of the two.
Exercise 82: Plotting a Correlated Matrix
In this exercise, we will compute the correlation between a pair of variables and draw a correlation plot using the corrplot package.
Perform the following steps to complete the exercise:
Import the required libraries using the following command:
library(mlbench) library(caret)
The output is...