2.3.4 Ridge Regression Model
Ridge regression33, is a dedicated to total linear biased estimation of regression data analysis method, is in essence a kind of improved least squares estimation method, by giving up the unbiasedness of least-square method, in order to lose some information, to reduce the accuracy at the expense of regression coefficient is more practical, more reliable regression method, the fitting of pathological data than the least square method. Ridge regression model is a widely used model34. Ridge regression is to artificially add a non-negative factor k to the main diagonal elements of the information matrix composed of independent variables, so that the matrix determinant is not zero, so as to reduce the error of regression coefficient estimation, improve the estimation accuracy and the model stationarity.Ridge regression can repair ill-conditioned matrix and achieve better results. Its implified diagram is shown in Figure 7.