The component estimating function is \(\delta(b)=\frac{\dot{\mu}(x^T b)}{\mu(x^T b)[1-\mu(x^T b)]}[y-\mu(x^T b)]w x\). Here, \(\mu\) is the inverse of the link function. In "logit" case, it can be simplified.
est_fun(y, X, b, link, w = NULL)numeric A vector of pseudo observations.
matrix The pairwise difference of original design matrix.
numeric The PIM parameter.
character The link function: "logit", or "probit".
numeric The weights, default is NULL.
matrix A n^2 x p matrix contains the n^2 transposed estimating fun.