Exercise 3. (5 points) For this exercise, the only extra package allowed is ISLR2. The datset

Boston will be used throughout the exercise and is accessible through the ISLR2 package.

We are interested in predicting medv given a single input variable from among the three

variables crim, rm and ptratio. We ran three regressions in R using each one of the three

variables as a single input variable. Then, we computed the average squared error between our

predictions for medv and the actual observed values of medv over the whole sample. We got

the following figure.

Inse

70

09

20

no

43

crim

mm

Index

Figure 4. Model Comparison

a) Write the R code to reproduce that plot. Explain the details of any calculations required.

b) Based on this plot, does it make sense to say that the best model among the three is

the one that uses rm as input? Or should it be the one that uses crim instead? Or

something else?