xrows(variates [,factors]), variates and factors REAL matrices or vectors or NULL. |

xrows(Variates, Factors) computes rows of an X-variable (design) matrix corresponding to variate values in Variates and factor levels in Factors, using the information saved by the preceding GLM command. In the following, let nv = number of variates in the model and nf = the number of factors in the model. Variates should be either a REAL vector with length(Variates) = nv, or a REAL matrix with ncols(Variates) = nv. when nv = 0, Variates should be NULL. Factors should be either a REAL vector with length(Factors) = nf or a REAL matrix with ncols(Factors) = nf. All the elements of Factors should be positive integers not exceeding the maximim level for each factor. When nf = 0, Factors should be NULL or should be omitted entirely. Let nrowv be 1 if Variates is a vector and nrows(Variates) otherwise, and let nrowf be 1 if Factors is a vector and nrows(Factors) otherwise. Then if nrowv != nrowf, you must have either nrowv = 1 or nrowf = 1 and the single row of Variates or Factors is used for every row of the output. The result is a REAL matrix with max(nrowv, nrowf) rows. Each row consists of the values of the X-variables (design matrix) corresponding to that row of Variates and Factors. Examples: After anova("y=x1+x2+a*b"), xrows(hconcat(x1,x2), hconcat(a,b)) is equivalent to xvariables(). After regress("y=x1+x2"); xrows(x0) %*% COEF is equivalent to regpred(x0,seest:F,sepred:F) where x0 is a matrix of values for x1 and x2. See also topics xvariables(), modelinfo(), regpred(), glmpred(), 'glm'.

Gary Oehlert 2003-01-15