least-squares regression

Statistical technique for estimating changes in a dependent variable (such as expenditure on food) which is in linear relationship with one or more independent variables (such as household income, size of the household, dietary needs, etc.). Based on fitting a straight line (called regression line or line of best fit) to the observed data (plotted as a scatter diagram) this technique aims to derive a good relationship (the best fit) that may be used to predict future values of one variable when the value of the other is known. It is so named because, in its computation, the sum of the squared deviations of the computed (future) values from the observed (past) values of the variables is minimized. Devised by the French mathematician Adrien-Marie Legendre (1752-1833), this technique is applicable to single line functions with any number of independent variables and, under certain assumptions, is the best statistical estimator.