The \(p\)-value and the standard error of estimated coefficient are not provided for tensor predictor regression since they depend on \(\widehat{\mathrm{cov}}^{-1}\{\mathrm{vec}(\mathbf{X})\}\) which is unavailable due to the ultra-high dimension of \(\mathrm{vec}(\mathbf{X})\).
print.summary.Tenv gives a more readable format of call, sample size, dimensions of datasets, mse. And if the object
is returned from TRR.fit
, then p-val
and se
are also returned.