pffr() for generalized data

[alexvolkmann]

Hey,

there is some problem with generalized functions in pffr() when the family argument is not specified as for example "poisson" but as "poisson()". Slightly adapted example from help:

data2 <- pffrSim(scenario="all", n=200) t <- attr(data2, "yindex") s <- attr(data2, "xindex") data2$Y_obs <- I(matrix(rpois(n = 200*60, lambda = exp(c(data2$Y))),nrow = 200, ncol = 60)) m2 <- pffr(Y_obs ~ ff(X1, xind=s) + xlin + c(te(xte1, xte2)) + s(xsmoo) + c(xconst), yind=t, data=data2, family = poisson())

This becomes a problem when you e.g. want to use different link function.

Best

[fabian-s]

oh no...

note to self:

if (!is.null(dots$family) && dots$family == "gaulss") { needs to check whether family was provided as a string or function call...