Linear regression, in simple terms, are used to predict/show a relationship between ONE response vairable (dependent) and one or more explanatory (predictor) vairables ; the amount dempends on whether you're doing multiple linear regression or simple linear regression. In MLR, often times building the model involves testing the variables through techniques like random forrest or principal component analysis. Linear regression is used very often in predictive analytics, you can build a model based on a store's previous locations and venues along with demographic info and be able to model/predict wherehighest revenue possible based on the selected predictor variables that make up the model.