Build a basic linear regression model on a CSV training data with python, and then evaluate the model's performance on the test data using mean squared error and R-squared.
Here we analyze a dataset with 471,744 instances or entries using Pandas, sklearn, matplotlib & Seaborn.