The recipes package contains a data preprocessor that can be used to avoid the potentially expensive formula methods as well as providing a richer set of data manipulation tools than base R can provide. This document uses version 1.0.10 of recipes.
In many cases, the preprocessing steps might contain quantities that require statistical estimation of parameters, such as
signal extraction using principal component analysis
imputation of missing values
transformations of individual variables (e.g. Box-Cox transformations)
It is critical that any complex preprocessing steps be contained inside of resampling so that the model performance estimates take into account the variability of these steps.
Before discussing how rsample can use recipes, let’s look at an example recipe for the Ames housing data.
An Example Recipe
For illustration, the Ames housing data will be used. There are sale prices of homes along with various other descriptors for the property:
data(ames, package = "modeldata")Suppose that we will again fit a simple regression model with the formula:
log10(Sale_Price) ~ Neighborhood + House_Style + Year_Sold + Lot_AreaThe distribution of the lot size is right-skewed:
library(ggplot2)
theme_set(theme_bw())
ggplot(ames, aes(x = Lot_Area)) +
geom_histogram(binwidth = 5000, col = "red", fill ="red", alpha = .5)
It might benefit the model if we estimate a transformation of the data using the Box-Cox procedure.
Also, note that the frequencies of the neighborhoods can vary:
ggplot(ames, aes(x = Neighborhood)) + geom_bar() + coord_flip() + xlab("")
When these are resampled, some neighborhoods will not be included in
the test set and this will result in a column of dummy variables with
zero entries. The same is true for the House_Style
variable. We might want to collapse rarely occurring values into “other”
categories.
To define the design matrix, an initial recipe is created:
library(recipes)
# Apply log10 transformation outside the recipe
# https://www.tmwr.org/recipes.html#skip-equals-true
ames <-
ames %>%
mutate(Sale_Price = log10(Sale_Price))
rec <-
recipe(Sale_Price ~ Neighborhood + House_Style + Year_Sold + Lot_Area,
data = ames) %>%
# Collapse rarely occurring jobs into "other"
step_other(Neighborhood, House_Style, threshold = 0.05) %>%
# Dummy variables on the qualitative predictors
step_dummy(all_nominal()) %>%
# Unskew a predictor
step_BoxCox(Lot_Area) %>%
# Normalize
step_center(all_predictors()) %>%
step_scale(all_predictors())
recThis recreates the work that the formula method traditionally uses with the additional steps.
While the original data object ames is used in the call,
it is only used to define the variables and their characteristics so a
single recipe is valid across all resampled versions of the data. The
recipe can be estimated on the analysis component of the resample.
If we execute the recipe on the entire data set:
rec_training_set <- prep(rec, training = ames)
rec_training_set## ## ── Recipe ────────────────────────────────────────────────────────────────## ## ── Inputs## Number of variables by role## outcome: 1
## predictor: 4## ## ── Training information## Training data contained 2930 data points and no incomplete rows.## ## ── Operations## • Collapsing factor levels for: Neighborhood and House_Style | Trained## • Dummy variables from: Neighborhood and House_Style | Trained## • Box-Cox transformation on: Lot_Area | Trained## • Centering for: Year_Sold and Lot_Area, ... | Trained## • Scaling for: Year_Sold and Lot_Area, ... | TrainedTo get the values of the data, the bake function can be
used:
# By default, the selector `everything()` is used to
# return all the variables. Other selectors can be used too.
bake(rec_training_set, new_data = head(ames))## # A tibble: 6 × 14
## Year_Sold Lot_Area Sale_Price Neighborhood_College_Creek
## <dbl> <dbl> <dbl> <dbl>
## 1 1.68 2.70 5.33 -0.317
## 2 1.68 0.506 5.02 -0.317
## 3 1.68 0.930 5.24 -0.317
## 4 1.68 0.423 5.39 -0.317
## 5 1.68 0.865 5.28 -0.317
## 6 1.68 0.197 5.29 -0.317
## # ℹ 10 more variables: Neighborhood_Old_Town <dbl>,
## # Neighborhood_Edwards <dbl>, Neighborhood_Somerset <dbl>,
## # Neighborhood_Northridge_Heights <dbl>, Neighborhood_Gilbert <dbl>,
## # Neighborhood_Sawyer <dbl>, Neighborhood_other <dbl>,
## # House_Style_One_Story <dbl>, House_Style_Two_Story <dbl>,
## # House_Style_other <dbl>Note that there are fewer dummy variables for
Neighborhood and House_Style than in the
data.
Also, the above code using prep() benefits from the
default argument of retain = TRUE, which keeps the
processed version of the data set so that we don’t have to reapply the
steps to extract the processed values. For the data used to train the
recipe, we would have used:
## # A tibble: 6 × 14
## Year_Sold Lot_Area Sale_Price Neighborhood_College_Creek
## <dbl> <dbl> <dbl> <dbl>
## 1 1.68 2.70 5.33 -0.317
## 2 1.68 0.506 5.02 -0.317
## 3 1.68 0.930 5.24 -0.317
## 4 1.68 0.423 5.39 -0.317
## 5 1.68 0.865 5.28 -0.317
## 6 1.68 0.197 5.29 -0.317
## # ℹ 10 more variables: Neighborhood_Old_Town <dbl>,
## # Neighborhood_Edwards <dbl>, Neighborhood_Somerset <dbl>,
## # Neighborhood_Northridge_Heights <dbl>, Neighborhood_Gilbert <dbl>,
## # Neighborhood_Sawyer <dbl>, Neighborhood_other <dbl>,
## # House_Style_One_Story <dbl>, House_Style_Two_Story <dbl>,
## # House_Style_other <dbl>The next section will explore recipes and bootstrap resampling for modeling:
library(rsample)
set.seed(7712)
bt_samples <- bootstraps(ames)
bt_samples## # Bootstrap sampling
## # A tibble: 25 × 2
## splits id
## <list> <chr>
## 1 <split [2930/1095]> Bootstrap01
## 2 <split [2930/1082]> Bootstrap02
## 3 <split [2930/1072]> Bootstrap03
## 4 <split [2930/1077]> Bootstrap04
## 5 <split [2930/1074]> Bootstrap05
## 6 <split [2930/1088]> Bootstrap06
## 7 <split [2930/1076]> Bootstrap07
## 8 <split [2930/1068]> Bootstrap08
## 9 <split [2930/1062]> Bootstrap09
## 10 <split [2930/1064]> Bootstrap10
## # ℹ 15 more rows
bt_samples$splits[[1]]## <Analysis/Assess/Total>
## <2930/1095/2930>Working with Resamples
We can add a recipe column to the tibble. recipes has a convenience
function called prepper() that can be used to call
prep() but has the split object as the first argument (for
easier purrring):
## # Bootstrap sampling
## # A tibble: 25 × 3
## splits id recipes
## <list> <chr> <list>
## 1 <split [2930/1095]> Bootstrap01 <recipe>
## 2 <split [2930/1082]> Bootstrap02 <recipe>
## 3 <split [2930/1072]> Bootstrap03 <recipe>
## 4 <split [2930/1077]> Bootstrap04 <recipe>
## 5 <split [2930/1074]> Bootstrap05 <recipe>
## 6 <split [2930/1088]> Bootstrap06 <recipe>
## 7 <split [2930/1076]> Bootstrap07 <recipe>
## 8 <split [2930/1068]> Bootstrap08 <recipe>
## 9 <split [2930/1062]> Bootstrap09 <recipe>
## 10 <split [2930/1064]> Bootstrap10 <recipe>
## # ℹ 15 more rows
bt_samples$recipes[[1]]## ## ── Recipe ────────────────────────────────────────────────────────────────## ## ── Inputs## Number of variables by role## outcome: 1
## predictor: 4## ## ── Training information## Training data contained 2930 data points and no incomplete rows.## ## ── Operations## • Collapsing factor levels for: Neighborhood and House_Style | Trained## • Dummy variables from: Neighborhood and House_Style | Trained## • Box-Cox transformation on: Lot_Area | Trained## • Centering for: Year_Sold and Lot_Area, ... | Trained## • Scaling for: Year_Sold and Lot_Area, ... | TrainedNow, to fit the model, the fit function only needs the recipe as
input. This is because the above code implicitly used the
retain = TRUE option in prep(). Otherwise, the
split objects would also be needed to bake() the recipe (as
it will in the prediction function below).
fit_lm <- function(rec_obj, ...)
lm(..., data = bake(rec_obj, new_data = NULL, everything()))
bt_samples$lm_mod <-
map(
bt_samples$recipes,
fit_lm,
Sale_Price ~ .
)
bt_samples## # Bootstrap sampling
## # A tibble: 25 × 4
## splits id recipes lm_mod
## <list> <chr> <list> <list>
## 1 <split [2930/1095]> Bootstrap01 <recipe> <lm>
## 2 <split [2930/1082]> Bootstrap02 <recipe> <lm>
## 3 <split [2930/1072]> Bootstrap03 <recipe> <lm>
## 4 <split [2930/1077]> Bootstrap04 <recipe> <lm>
## 5 <split [2930/1074]> Bootstrap05 <recipe> <lm>
## 6 <split [2930/1088]> Bootstrap06 <recipe> <lm>
## 7 <split [2930/1076]> Bootstrap07 <recipe> <lm>
## 8 <split [2930/1068]> Bootstrap08 <recipe> <lm>
## 9 <split [2930/1062]> Bootstrap09 <recipe> <lm>
## 10 <split [2930/1064]> Bootstrap10 <recipe> <lm>
## # ℹ 15 more rowsTo get predictions, the function needs three arguments: the splits
(to get the assessment data), the recipe (to process them), and the
model. To iterate over these, the function purrr::pmap() is
used:
pred_lm <- function(split_obj, rec_obj, model_obj, ...) {
mod_data <- bake(
rec_obj,
new_data = assessment(split_obj),
all_predictors(),
all_outcomes()
)
out <- mod_data %>% select(Sale_Price)
out$predicted <- predict(model_obj, newdata = mod_data %>% select(-Sale_Price))
out
}
bt_samples$pred <-
pmap(
lst(
split_obj = bt_samples$splits,
rec_obj = bt_samples$recipes,
model_obj = bt_samples$lm_mod
),
pred_lm
)
bt_samples## # Bootstrap sampling
## # A tibble: 25 × 5
## splits id recipes lm_mod pred
## <list> <chr> <list> <list> <list>
## 1 <split [2930/1095]> Bootstrap01 <recipe> <lm> <tibble [1,095 × 2]>
## 2 <split [2930/1082]> Bootstrap02 <recipe> <lm> <tibble [1,082 × 2]>
## 3 <split [2930/1072]> Bootstrap03 <recipe> <lm> <tibble [1,072 × 2]>
## 4 <split [2930/1077]> Bootstrap04 <recipe> <lm> <tibble [1,077 × 2]>
## 5 <split [2930/1074]> Bootstrap05 <recipe> <lm> <tibble [1,074 × 2]>
## 6 <split [2930/1088]> Bootstrap06 <recipe> <lm> <tibble [1,088 × 2]>
## 7 <split [2930/1076]> Bootstrap07 <recipe> <lm> <tibble [1,076 × 2]>
## 8 <split [2930/1068]> Bootstrap08 <recipe> <lm> <tibble [1,068 × 2]>
## 9 <split [2930/1062]> Bootstrap09 <recipe> <lm> <tibble [1,062 × 2]>
## 10 <split [2930/1064]> Bootstrap10 <recipe> <lm> <tibble [1,064 × 2]>
## # ℹ 15 more rowsCalculating the RMSE:
library(yardstick)
results <- map(bt_samples$pred, rmse, Sale_Price, predicted) %>% list_rbind()
results## # A tibble: 25 × 3
## .metric .estimator .estimate
## <chr> <chr> <dbl>
## 1 rmse standard 0.132
## 2 rmse standard 0.128
## 3 rmse standard 0.129
## 4 rmse standard 0.123
## 5 rmse standard 0.125
## 6 rmse standard 0.140
## 7 rmse standard 0.129
## 8 rmse standard 0.130
## 9 rmse standard 0.122
## 10 rmse standard 0.127
## # ℹ 15 more rows
mean(results$.estimate)## [1] 0.129