pboesu / ewp Goto Github PK

When trying to scale a variable e.g. scale(min_first_egg) within a model, I get "Error in scale(min_first_egg): object 'min_first_egg' not found.

This didn't happen previously and doesn't happen with the same code in a GLM.

fit_min <- ewp_reg(max_num_eggs ~ scale(min_first_egg, scale = F), data = wilwa_trial)
summary(fit_min)

Model3_min<-glm(max_num_eggs~scale(min_first_egg, scale = F), family = quasipoisson, data = wilwa_trial)
summary(Model3_min)

The data are the same that I sent previously.

Set lower limit to estimated parscale avoid division by (near) zero.

The PMF calculation requires evaluation of an infinite sum of weights, which is currently approximate using a hard coded upper limit for the sum. ewp_reg currently issues a warning when that limit is approached, but lower level functions do not currently handle this defensively.
Ideally all relevant functions should adapt the summation limit conditional on lambda and beta2, with warnings issued for counts above some lambda to point out the increasing computational effort as the summation limit increases.

Full error message:
Error in fetch(key) :
lazy-load database 'C:/Users/jerem/Dropbox/PC/Documents/R/win-library/4.1/ewp/help/ewp.rdb' is corrupt

This occurred whilst trying to run my histogram plotting function using the model outputs from ewp - particularly the null model. Data were wilwa_trial data.

May be related to other error:

Error: Expecting a single value: [extent=10].
3. stop(structure(list(message = "Expecting a single value: [extent=10].",
call = NULL, cppstack = NULL), class = c("Rcpp::not_compatible",
"C++Error", "error", "condition")))
2. dewp3_cpp(seq(1, 10), lambda = model$coefficients[[1]], beta1 = fit_null$coefficients[[2]],
beta2 = fit_null$coefficients[[3]])

1. hist.ewp.null(bullf$max_num_eggs[bullf$year == 1960], fit_null)

Second error has been solved by providing a singular value to x