The function takes an object created by x_Obj as input and add
response values. Further initial computations for prediction and testing is
made.
xy_Obj(xObj, Y)ffModelObj(
xObj,
Y,
modelMatrix,
modelTerms,
model,
xlev,
scaleY,
scaleX,
centerX,
isIntercept,
returnY = FALSE,
returnYhat = FALSE,
returnYhatStd = FALSE
)
object created by x_Obj
response matrix
Model matrix (output from model.matrix) to be included in output.
Model terms (model frame attribute) to be included in output.
Values used to scale Y (see stdize) to be included in output.
Values used to scale the model matrix (see stdize) to be included in output.
Values used to center the model matrix (see stdize) to be included in output.
A logical (whether model has intercept) to be included in output.
Matrix Y (as input) in output when TRUE.
Matrix Yhat of fitted values corresponding to Y in output when TRUE.
Standard errors, YhatStd, in output when TRUE.
A list with components
same as input
same as input
equals sum(Y^2)
equals
sum((center(Y))^2). That is, the total sum of squares summed over all
responses.
Sums of squares summed over all responses.
Output from linregEst where xObj$D_om is the
regressor matrix.
fitted values
standard deviations of fitted values
mean square error of each response
Error observations that can be used in multivariate testing
Hypothesis observations that can be used in multivariate testing
Traditionally, sums of squares and cross-products (SSC) is the multivariate
generalisation of sums of squares. When there is a large number of responses
this representation is inefficient and therefore linear combinations of
observations (Langsrud, 2002) is stored instead, such as errorObs.
The corresponding SSC matrix can be obtained by
t(errorObs)%*%errorObs. When there is a large number of observations
the errorObs representation is also inefficient, but it these cases it is
possible to chose a representation with several zero rows. Then, errorObs is
stored as a two-component list: A matrix containing the nonzero rows of
errorObs and an integer representing the degrees of freedom for error
(number of rows in the full errorObs matrix).
Langsrud, <U+00D8>. (2002) 50-50 Multivariate Analysis of Variance for Collinear Responses. The Statistician, 51, 305--317.