​

​
Linear regression is used to see how well an outcome (’dependent’) variable can be predicted by another (’independent’) variable. Even if there is a high correlation overall, it’s important to look at the graph to see how far off the mark a prediction can be before using it to guide decisions (see red dots).