Add better error message to h2o.ensemble for misspecified custom learner wrappers

When a user tries to make a custom wrapper by wrapping h2o.glm instead of h2o.glm.wrapper, for example, the error message is not informative. We should do a check to make sure a wrapper function is correct and stop the user if it's not. Example below.

The problem is that there is a keep_cross_validation_folds argument that's trying to be passed to a learner wrapper and it no longer passes through silently (this is a typo of the real argument name).

Adding this custom learner wrapper triggers the bug: {code} h2o.glm.lbfgs <- function(..., solver = "L_BFGS") { h2o.glm(..., solver = solver)} {code}

{code} Error in h2o.glm(..., solver = solver) : unused argument (keep_cross_validation_folds = TRUE) 9 h2o.glm(..., solver = solver) 8 match.fun(learner[l])(y = y, x = x, training_frame = training_frame, validation_frame = NULL, family = family, fold_column = fold_column, keep_cross_validation_folds = TRUE) 7 .fitFun(l, y, xcols, training_frame, validation_frame, family, learner, seed, fold_column) 6 system.time(fit <- .fitFun(l, y, xcols, training_frame, validation_frame, family, learner, seed, fold_column), gcFirst = FALSE) 5 FUN(X[[i]], ...) 4 lapply(X = X, FUN = FUN, ...) 3 sapply(X = 1:L, FUN = .fitWrapper, y = y, xcols = x, training_frame = training_frame, validation_frame = NULL, family = family, learner = learner, seed = seed, fold_column = "fold_id", simplify = FALSE) 2 .make_Z(x = x, y = y, training_frame = training_frame, family = family, learner = learner, parallel = parallel, seed = seed, V = V, L = L, idxs = idxs, metalearner_type = metalearner_type) 1 h2o.ensemble(x = x, y = y, training_frame = train, family = family, learner = learner, metalearner = metalearner, cvControl = list(V = 5, shuffle = TRUE)) Timing stopped at: 0.003 0 0.002 {code}

h2oai / h2o-3

Add better error message to h2o.ensemble for misspecified custom learner wrappers #9969