Linear regression is a population model that relates a target variable to one or multiple regressors , which can be linear or non-linear function(s) of independent variables, through an equation that is linear in parameters and the error : + + . The target and independent variables, and hence the parameters , all belong to the population and not a specific sample (hence the name population model). The subscripts and refer to different observations and independent variables, respectively, and the error (not to be confused with residual) represents the variations (from one observation to the other) in that are not explainable with variations in the s. The actual value of the population parameters will be known only if we measure the entire population, which is almost never possible. Therefore, we have no choice but to estimate these parameters somehow and use the estimated values in our population model.