regpred(vals [, silent:T]), vals a REAL vector or matrix. |

regpred(vals) computes the fitted (predicted) value, the standard error of estimation, and the standard error of prediction for the current regression model when the X variables have values given by the REAL vector or matrix vals. When there is only 1 variate in the model, vals is a scalar or a vector and the estimates and standard errors are computed for each element of vals. When the number of variates in the model is nvars > 1, vals must either be a vector of length nvars containing data for a single case, or a m by nvars matrix containing data for m cases. In the latter case, each component of the result is a vector of length m. For example, after regress("y=x1+x2+x3"), regpred(hconcat(x1,x2,x3)) computes the predicted values and standard errors for all cases in the data set. The result is a structure with components 'estimate', 'SEest', and 'SEpred'. When the error degrees of freedom are 0, all standard errors are set to MISSING. Caution: After anova(), manova() and regress(), standard errors are computed using the final error mean square in the model. This may not be appropriate with mixed models, including split plot designs. regpred(vals, silent:T) does the same except certain advisory messages are suppressed. The default value of 'silent' is False unless the value of option' 'warnings' is False. You can also use keyword phrases estimate:F, seest:F, sepred:F and n:N. See glmpred() for details. You can use regpred() after any GLM command as long as there are no factors in the model. The output has no SEpred component except after regress(), anova() or manova() or their weighted versions. After anova(), manova() and regress(), regpred(vals) is equivalent to glmpred(vals,sepred:T). After other GLM commands, regpred(vals) is equivalent to glmpred(vals). See also topics regress(), anova(), glmpred(), predtable(), glmtable(), modelinfo(), popmodel(), pushmodel(), 'glm', yhat().

Gary Oehlert 2003-01-15