Modeltime.resample - 为什么 modeltime_fit_resamples() 不接受“未拟合的”modeltime_table 作为输入？

如何解决Modeltime.resample - 为什么 modeltime_fit_resamples() 不接受“未拟合的”modeltime_table 作为输入？

我在这里的主要目标是更好地理解使用 modeltime_fit_resamples() 而不是传统的 fit_resamples() 做事方式的优势。

与fit_resamples()函数不同，modeltime_fit_resamples的输入 必须是一个合适的“Modeltime Table”，这听起来像是在浪费资源。如果此函数不使用该信息，为什么要在运行 modeltime_fit_resamples() 之前拟合模型？也许这是一个愚蠢的问题，所以我提前道歉。

以下可重现的示例：

    library(modeltime)
    library(modeltime.resample)
    library(tidyverse)
    library(timetk)
    library(earth)
    
    
    # Set Up Model Workflow and Workflow fit ----
        model_spec_mars <- mars(mode = "regression") %>%
          set_engine("earth") 
        
        recipe_spec <- recipe(value ~ date,data = training(splits)) %>%
          step_date(date,features = "month",ordinal = FALSE) %>%
          step_mutate(date_num = as.numeric(date)) %>%
          step_normalize(date_num) %>%
          step_rm(date)
        
        #Workflow
        wflw_mars <- workflow() %>%
          add_recipe(recipe_spec) %>%
          add_model(model_spec_mars) 
        
        #Fitted Workflow
        splits <- initial_time_split(m750,prop = 0.9)
        wflw_mars_fit <- wflw_mars %>% 
          fit(training(splits))
    
    # Resamples --------
        
        resamples_tscv <- time_series_cv(
          data        = m750,assess      = "2 years",initial     = "5 years",skip        = "2 years",slice_limit = 4
        )
        
        resamples_tscv
        
    # fit_resamples Example --------
    
        fit_output <- 
          wflw_mars %>% 
          fit_resamples(
            resamples_tscv,control=control_resamples(verbose=T,save_pred=T)
          )
    
    # modeltime_fit_resamples (does not work )--------    
        
    fit_output <-
          wflw_mars %>% 
          modeltime_table() %>% 
          modeltime_fit_resamples(
            resamples_tscv,save_pred=T)
      )
    
        
    # modeltime_fit_resamples ( Works )--------    
        
        wflw_mars %>% 
          fit(training(splits)) %>% 
          modeltime_table() %>% 
          modeltime_fit_resamples(
            resamples_tscv,save_pred=T)
          )
        
        #it works,but why would I fit the model if modeltime_fit_resamples does not use that information ?