The least squares criterion is the rule that selects, from all possible lines or curves, the one that minimises the sum of the squared residuals, where a residual is the vertical gap between an observed value and the value the line predicts. Squaring makes every gap positive and gives large gaps more weight than small ones. It is the basis of ordinary least squares (OLS), the standard method behind linear regression and much of econometrics. For a simple line, the criterion gives a slope and an intercept that can be calculated directly from the data.
OLS gives good estimates under assumptions that include a linear relationship, errors that average zero and have constant variance (the opposite of heteroskedasticity), and errors that are not correlated with each other. Because gaps are squared, a few extreme observations can pull the line sharply, so the method is sensitive to outliers. The fit is usually summarised by the coefficient of determination, and the wider method is described under regression analysis.
Ref 1100001970