I took a course on Udacity and the focus was on using the Predictive tools, examining the output of the reports, and deciding which variable to use. After I came up with a solution I wanted to check against the solution, which starts with the Data Investigation palette. My list of predictor variables was very close to the solution file.
Also, I compared three different models and used the Forest Model, which had better correlation and error measures.
I had to change the Linear Regression tool to use the 1.0 version in order to get an output that matched the solution file.