You may be asking yourself whether you will ever have just one predictor variable in the real world. That is indeed a fair question and certainly a very rare case (time series can be a common exception). Most likely, several, if not many, predictor variables or features--as they are affectionately termed in machine learning--will have to be included in your model. And with that, let's move on to multivariate linear regression and a new business case.
In keeping with the water conservation/prediction theme, let's look at another dataset in the alr3
package, appropriately named water
. During the writing of the first edition of this book, the severe drought in Southern California caused much alarm. Even the Governor, Jerry Brown, began to take action with a call to citizens to reduce water usage by 20 percent. For this exercise, let's say we have been commissioned by the state of California to predict water availability. The data provided...