Multi predict?

[DavisVaughan]

I don't think it will immediately be required by tune but would still be useful

[jaredlander]

Just experiencing this. I built a glmnet model using {workflows} and multi_predict() does not exist for the workflow object. Though it does exist for the fitted model stored inside the workflow object.

rec <- recipe(y ~ x, data) %>%
    step_(...)
lm_spec <- linear_reg(mode='regression', mixture=1) %>% 
    set_engine('glmnet')

flow <- workflow() %>% 
    add_recipe(rec) %>% 
    add_model(lm_spec)

mod <- flow %>% fit(data=data)

has_multi_predict(mod)
[1] FALSE
has_multi_predict(mod$fit$fit)
[1] TRUE

So I am mapping over different values of lambda rather than calling multi_predict().