yhat() or yhat(Model [,T]) |

yhat(), with no argument, computes a REAL matrix of various quantities useful in making predictions from a regression or analysis of variance model. yhat() uses side effect variables RESIDUALS, HII, etc. produced by the most recent GLM (generalized linear or linear model) command such as regress() or anova(). If weights were supplied, it also uses the result of modelinfo(weights:T). NOTE: If the most recent GLM was not linear, that is, not regress(), anova(), manova() or their weighted variants, only the first two columns computed by yhat() are meaningful. It is an error if any of the needed side effect variables do not exist. yhat(Model) first executes manova(Model, silent:T) to compute the side effect variables and then computes its usual output. Model should be a CHARACTER variable or string specifying a linear model. Any factors in the model will be treated as factors. yhat(Model,T) does the same, except any factors in the model are treated as variates. Each row of the result corresponds to a case. When the dependent variable Y is univariate (has one column), the result has the following 5 columns: Col. 1 Y = observed response Col. 2 Yhat = predicted or fitted value computed using all data Col. 3 Predictive residuals = Y - (Yhat computed excluding the case) Col. 4 SE[Yhat] = estimated standard error of Yhat as estimate of E[Y | x] Col. 5 SE[pred] = estimated s.e. of prediction error When the all the independent variables in the model "y=x1+x2+x3+...+xk" are variates and not factors, columns 2, 4, and 5 of the output correspond to components 'estimate', 'SEest', and 'SEpred" in the output of regpred(hconcat(x1,x2,...)) following regress("y=x1+x2+...+xk"). See regpred(). When Y is multivariate, with p columns, there are 5*p columns in groups of p -- the p columns of Y, the p columns of YHat, and so on. If a case has missing values, most entries will be MISSING and there are no useful numbers. Example: Cmd> yhat("y=x1+x2+x3+x4") The output of yhat() is modelled on output printed by the yhat() command in program Multreg. yhat() is implemented as a pre-defined macro. See also topics regpred(), 'glm', resid().

Gary Oehlert 2003-01-15