Tables for engine specific params in model docs

[juliasilge]

Addresses #211. This PR implements a table in each individual model .Rd file to show the mapping from the parsnip parameters to the engine parameters.

Still TODO is the defaults for these parameters but that is a WHOLE THING (maybe not something that can be automated) so I plan to do that in another PR in the hopefully near future. If this piece is good as is, let's merge it.

[topepo]

Looks good. I agree that the default arguments should be a second stage.

[topepo]

For future PRs... this is a little kludgy but we could add some code to the _data files that define the methods to catalog their default values:

library(tidymodels)
#> ── Attaching packages ───────────────────────────── tidymodels 0.1.0 ──
#> ✓ broom     0.5.4          ✓ recipes   0.1.10    
#> ✓ dials     0.0.4.9000     ✓ rsample   0.0.5.9000
#> ✓ dplyr     0.8.5          ✓ tibble    2.1.3     
#> ✓ ggplot2   3.3.0          ✓ tune      0.0.1.9000
#> ✓ infer     0.5.1          ✓ workflows 0.1.0     
#> ✓ parsnip   0.0.5          ✓ yardstick 0.0.5     
#> ✓ purrr     0.3.3
#> ── Conflicts ──────────────────────────────── tidymodels_conflicts() ──
#> x purrr::discard()  masks scales::discard()
#> x dplyr::filter()   masks stats::filter()
#> x dplyr::lag()      masks stats::lag()
#> x ggplot2::margin() masks dials::margin()
#> x recipes::step()   masks stats::step()
library(rlang)
#> 
#> Attaching package: 'rlang'
#> The following objects are masked from 'package:purrr':
#> 
#>     %@%, as_function, flatten, flatten_chr, flatten_dbl, flatten_int,
#>     flatten_lgl, flatten_raw, invoke, list_along, modify, prepend,
#>     splice


get_arg <- function(ns, f, arg) {
  args <- formals(getFromNamespace(f, ns))
  args <- args %>% as.list() 
  as.character(args[[arg]])
}

# Make the defaults character because there are cases where we will write something
# eg glmnet::glmnet would have "lambda (all)" or something similar
dt_defaults <- 
  tibble::tribble(
    ~model,         ~engine,                 ~original,  ~default,
    "decision_tree", "rpart",               "maxdepth", get_arg("rpart", "rpart.control", "maxdepth"),
    "decision_tree", "rpart",               "minsplit", get_arg("rpart", "rpart.control", "minsplit"),
    "decision_tree", "rpart",                     "cp", get_arg("rpart", "rpart.control", "cp"),
    "decision_tree",  "C5.0",               "minCases", get_arg("C50", "C5.0Control", "minCases"),
    "decision_tree", "spark",              "max_depth", get_arg("sparklyr", "ml_decision_tree", "max_depth"),
    "decision_tree", "spark", "min_instances_per_node", get_arg("sparklyr", "ml_decision_tree", "min_instances_per_node"),
  )

# emulating convert_args("decision_tree")
model_name <- "decision_tree"

envir <- get_model_env()

args <-
  ls(envir) %>%
  tibble::tibble(name = .) %>%
  dplyr::filter(grepl("args", name)) %>%
  dplyr::mutate(model = sub("_args", "", name),
                args  = purrr::map(name, ~envir[[.x]])) %>%
  dplyr::filter(grepl(model_name, model)) %>%
  tidyr::unnest(args) %>%
  dplyr::select(model:original) %>% 
  full_join(dt_defaults) %>% 
  mutate(original = paste0(original, " (", default, ")")) %>% 
  select(-default)
#> Joining, by = c("model", "engine", "original")

convert_df <- args %>%
  dplyr::select(-model) %>%
  tidyr::pivot_wider(names_from = engine, values_from = original)

convert_df %>%
  knitr::kable(col.names = paste0("**", colnames(convert_df), "**"))

parsnip	rpart	C5.0	spark
tree_depth	maxdepth (30)	NA	max_depth (5)
min_n	minsplit (20)	minCases (2)	min_instances_per_node (1)
cost_complexity	cp (0.01)	NA	NA

^{Created on 2020-03-24 by the reprex package (v0.3.0)}

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.