Updated for workshop at RStudio::conf 2020; added readme

README.md

@@ -0,0 +1,11 @@
+\# RStudio Training Materials for tidymodels
+
+These directories contain training materials that we've presented for the `tidymodels` packages. We keep this repo up-to-date with the most recent content.
+
+They are organized by the length of training. The current set of materials:
+
+* `two day`: from RStudio::conf 2020
+
+* `one day`: 2019 Symposium on Data Science and Statistics (these are fairly out-of-date and don't include newer packages)
+
+* `half day`: 2019 R/Pharma conference (these are fairly out-of-date and don't include newer packages)

two day/Chicago_grid.R

@@ -0,0 +1,68 @@
+library(tidymodels)
+library(tune)
+
+library(doMC)
+registerDoMC(cores = 8)
+
+# ------------------------------------------------------------------------------
+
+set.seed(7898)
+data_folds <- rolling_origin(Chicago, initial = 364 * 15, assess = 7 * 4, skip = 7 * 4, cumulative = FALSE)
+
+# ------------------------------------------------------------------------------
+
+library(stringr)
+us_hol <- 
+  timeDate::listHolidays() %>% 
+  str_subset("(^US)|(Easter)")
+
+chi_rec <-
+  recipe(ridership ~ ., data = Chicago) %>%
+  step_holiday(date, holidays = us_hol) %>%
+  step_date(date) %>%
+  step_rm(date) %>%
+  step_dummy(all_nominal()) %>%
+  step_zv(all_predictors())
+
+mars_rec <- 
+  chi_rec %>% 
+  step_normalize(one_of(!!stations)) %>% 
+  step_pca(one_of(!!stations), num_comp = tune("pca comps"))
+
+mars_mod <-  
+  mars(num_terms = tune("mars terms"), prod_degree = tune(), prune_method = "none") %>% 
+  set_engine("earth") %>% 
+  set_mode("regression")
+
+chi_wflow <-
+  workflow() %>%
+  add_recipe(mars_rec) %>%
+  add_model(mars_mod)
+
+chi_grid <- 
+  expand.grid(
+    `pca comps`  = 0:20,
+    prod_degree = 1:2,
+    `mars terms` = 2:100
+  )
+
+all_res <-
+  tune_grid(
+    chi_wflow,
+    resamples = data_folds,
+    grid = chi_grid,
+    control = control_grid(verbose = TRUE)
+  )
+
+print(all_res$.notes[[1]])
+
+complete_mars_grid <-
+  all_res %>% 
+  collect_metrics() %>% 
+  dplyr::filter(.metric == "rmse")
+
+save(complete_mars_grid, file = "complete_mars_grid.RData")
+
+q("no")
+
+