 Consider a few real world problems like predicting the price of a house, or determining a credit card score, or predicting the number of subscribers I'll have by 2020. All of these involve predicting a value and are hence regression type problems. Each one of these problems has a set of input variables called covariates that are used to predict an output response variable. Let's go by other names like features or independent variables and predictors. The relationship between these input and output variables are determined using a statistical learning method, specifically for regression in this case.