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 0.1.4 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:

library(AmesHousing)
ames <- make_ames()
names(ames)
##  [1] "MS_SubClass"        "MS_Zoning"          "Lot_Frontage"      
##  [4] "Lot_Area"           "Street"             "Alley"             
##  [7] "Lot_Shape"          "Land_Contour"       "Utilities"         
## [10] "Lot_Config"         "Land_Slope"         "Neighborhood"      
## [13] "Condition_1"        "Condition_2"        "Bldg_Type"         
## [16] "House_Style"        "Overall_Qual"       "Overall_Cond"      
## [19] "Year_Built"         "Year_Remod_Add"     "Roof_Style"        
## [22] "Roof_Matl"          "Exterior_1st"       "Exterior_2nd"      
## [25] "Mas_Vnr_Type"       "Mas_Vnr_Area"       "Exter_Qual"        
## [28] "Exter_Cond"         "Foundation"         "Bsmt_Qual"         
## [31] "Bsmt_Cond"          "Bsmt_Exposure"      "BsmtFin_Type_1"    
## [34] "BsmtFin_SF_1"       "BsmtFin_Type_2"     "BsmtFin_SF_2"      
## [37] "Bsmt_Unf_SF"        "Total_Bsmt_SF"      "Heating"           
## [40] "Heating_QC"         "Central_Air"        "Electrical"        
## [43] "First_Flr_SF"       "Second_Flr_SF"      "Low_Qual_Fin_SF"   
## [46] "Gr_Liv_Area"        "Bsmt_Full_Bath"     "Bsmt_Half_Bath"    
## [49] "Full_Bath"          "Half_Bath"          "Bedroom_AbvGr"     
## [52] "Kitchen_AbvGr"      "Kitchen_Qual"       "TotRms_AbvGrd"     
## [55] "Functional"         "Fireplaces"         "Fireplace_Qu"      
## [58] "Garage_Type"        "Garage_Finish"      "Garage_Cars"       
## [61] "Garage_Area"        "Garage_Qual"        "Garage_Cond"       
## [64] "Paved_Drive"        "Wood_Deck_SF"       "Open_Porch_SF"     
## [67] "Enclosed_Porch"     "Three_season_porch" "Screen_Porch"      
## [70] "Pool_Area"          "Pool_QC"            "Fence"             
## [73] "Misc_Feature"       "Misc_Val"           "Mo_Sold"           
## [76] "Year_Sold"          "Sale_Type"          "Sale_Condition"    
## [79] "Sale_Price"         "Longitude"          "Latitude"

Suppose that we will again fit a simple regression model with the formula:

log10(Sale_Price) ~ Neighborhood + House_Style + Year_Sold + Lot_Area

The 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 entires. 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:

## Data Recipe
## 
## Inputs:
## 
##       role #variables
##    outcome          1
##  predictor          4
## 
## Operations:
## 
## Log transformation on Sale_Price
## Collapsing factor levels for Neighborhood, House_Style
## Dummy variables from all_nominal()
## Box-Cox transformation on Lot_Area
## Centering for all_predictors()
## Scaling for all_predictors()

This 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:

## Data Recipe
## 
## Inputs:
## 
##       role #variables
##    outcome          1
##  predictor          4
## 
## Training data contained 2930 data points and no missing data.
## 
## Operations:
## 
## Log transformation on Sale_Price [trained]
## Collapsing factor levels for Neighborhood, House_Style [trained]
## Dummy variables from Neighborhood, House_Style [trained]
## Box-Cox transformation on Lot_Area [trained]
## Centering for Year_Sold, ... [trained]
## Scaling for Year_Sold, ... [trained]

To get the values of the data, the bake function can be used:

## # A tibble: 6 x 14
##   Lot_Area Year_Sold Sale_Price Neighborhood_Co… Neighborhood_Ol…
##      <dbl>     <dbl>      <dbl>            <dbl>            <dbl>
## 1    2.70       1.68       5.33           -0.317           -0.298
## 2    0.506      1.68       5.02           -0.317           -0.298
## 3    0.930      1.68       5.24           -0.317           -0.298
## 4    0.423      1.68       5.39           -0.317           -0.298
## 5    0.865      1.68       5.28           -0.317           -0.298
## 6    0.197      1.68       5.29           -0.317           -0.298
## # ... with 9 more variables: 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:

juice(rec_training_set) %>% head
## # A tibble: 6 x 14
##   Year_Sold Lot_Area Sale_Price Neighborhood_Co… Neighborhood_Ol…
##       <dbl>    <dbl>      <dbl>            <dbl>            <dbl>
## 1      1.68    2.70        5.33           -0.317           -0.298
## 2      1.68    0.506       5.02           -0.317           -0.298
## 3      1.68    0.930       5.24           -0.317           -0.298
## 4      1.68    0.423       5.39           -0.317           -0.298
## 5      1.68    0.865       5.28           -0.317           -0.298
## 6      1.68    0.197       5.29           -0.317           -0.298
## # ... with 9 more variables: 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:

## # Bootstrap sampling 
## # A tibble: 25 x 2
##    splits              id         
##    <list>              <chr>      
##  1 <split [2.9K/1.1K]> Bootstrap01
##  2 <split [2.9K/1.1K]> Bootstrap02
##  3 <split [2.9K/1.1K]> Bootstrap03
##  4 <split [2.9K/1K]>   Bootstrap04
##  5 <split [2.9K/1.1K]> Bootstrap05
##  6 <split [2.9K/1.1K]> Bootstrap06
##  7 <split [2.9K/1.1K]> Bootstrap07
##  8 <split [2.9K/1K]>   Bootstrap08
##  9 <split [2.9K/1.1K]> Bootstrap09
## 10 <split [2.9K/1.1K]> Bootstrap10
## # ... with 15 more rows
## <2930/1076/2930>

Working with Resamples

We can add a recipe column to the tibble. recipes has a connivence function called prepper that can be used to call prep but has the split object as the first argument (for easier purrring):

library(purrr)

bt_samples$recipes <- map(bt_samples$splits, prepper, recipe = rec)
bt_samples
## # Bootstrap sampling 
## # A tibble: 25 x 3
##    splits              id          recipes     
##    <list>              <chr>       <list>      
##  1 <split [2.9K/1.1K]> Bootstrap01 <S3: recipe>
##  2 <split [2.9K/1.1K]> Bootstrap02 <S3: recipe>
##  3 <split [2.9K/1.1K]> Bootstrap03 <S3: recipe>
##  4 <split [2.9K/1K]>   Bootstrap04 <S3: recipe>
##  5 <split [2.9K/1.1K]> Bootstrap05 <S3: recipe>
##  6 <split [2.9K/1.1K]> Bootstrap06 <S3: recipe>
##  7 <split [2.9K/1.1K]> Bootstrap07 <S3: recipe>
##  8 <split [2.9K/1K]>   Bootstrap08 <S3: recipe>
##  9 <split [2.9K/1.1K]> Bootstrap09 <S3: recipe>
## 10 <split [2.9K/1.1K]> Bootstrap10 <S3: recipe>
## # ... with 15 more rows
## Data Recipe
## 
## Inputs:
## 
##       role #variables
##    outcome          1
##  predictor          4
## 
## Training data contained 2930 data points and no missing data.
## 
## Operations:
## 
## Log transformation on Sale_Price [trained]
## Collapsing factor levels for Neighborhood, House_Style [trained]
## Dummy variables from Neighborhood, House_Style [trained]
## Box-Cox transformation on Lot_Area [trained]
## Centering for Year_Sold, ... [trained]
## Scaling for Year_Sold, ... [trained]

Now, 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).

## # Bootstrap sampling 
## # A tibble: 25 x 4
##    splits              id          recipes      lm_mod  
##    <list>              <chr>       <list>       <list>  
##  1 <split [2.9K/1.1K]> Bootstrap01 <S3: recipe> <S3: lm>
##  2 <split [2.9K/1.1K]> Bootstrap02 <S3: recipe> <S3: lm>
##  3 <split [2.9K/1.1K]> Bootstrap03 <S3: recipe> <S3: lm>
##  4 <split [2.9K/1K]>   Bootstrap04 <S3: recipe> <S3: lm>
##  5 <split [2.9K/1.1K]> Bootstrap05 <S3: recipe> <S3: lm>
##  6 <split [2.9K/1.1K]> Bootstrap06 <S3: recipe> <S3: lm>
##  7 <split [2.9K/1.1K]> Bootstrap07 <S3: recipe> <S3: lm>
##  8 <split [2.9K/1K]>   Bootstrap08 <S3: recipe> <S3: lm>
##  9 <split [2.9K/1.1K]> Bootstrap09 <S3: recipe> <S3: lm>
## 10 <split [2.9K/1.1K]> Bootstrap10 <S3: recipe> <S3: lm>
## # ... with 15 more rows

To 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:

## # Bootstrap sampling 
## # A tibble: 25 x 5
##    splits              id          recipes      lm_mod  pred              
##    <list>              <chr>       <list>       <list>  <list>            
##  1 <split [2.9K/1.1K]> Bootstrap01 <S3: recipe> <S3: l… <tibble [1,076 × …
##  2 <split [2.9K/1.1K]> Bootstrap02 <S3: recipe> <S3: l… <tibble [1,105 × …
##  3 <split [2.9K/1.1K]> Bootstrap03 <S3: recipe> <S3: l… <tibble [1,102 × …
##  4 <split [2.9K/1K]>   Bootstrap04 <S3: recipe> <S3: l… <tibble [1,015 × …
##  5 <split [2.9K/1.1K]> Bootstrap05 <S3: recipe> <S3: l… <tibble [1,070 × …
##  6 <split [2.9K/1.1K]> Bootstrap06 <S3: recipe> <S3: l… <tibble [1,058 × …
##  7 <split [2.9K/1.1K]> Bootstrap07 <S3: recipe> <S3: l… <tibble [1,081 × …
##  8 <split [2.9K/1K]>   Bootstrap08 <S3: recipe> <S3: l… <tibble [1,048 × …
##  9 <split [2.9K/1.1K]> Bootstrap09 <S3: recipe> <S3: l… <tibble [1,090 × …
## 10 <split [2.9K/1.1K]> Bootstrap10 <S3: recipe> <S3: l… <tibble [1,072 × …
## # ... with 15 more rows

Calculating the RMSE:

library(yardstick)
## For binary classification, the first factor level is assumed to be the event.
## Set the global option `yardstick.event_first` to `FALSE` to change this.
results <- map_dfr(bt_samples$pred, rmse, Sale_Price, predicted)
results
## # A tibble: 25 x 3
##    .metric .estimator .estimate
##    <chr>   <chr>          <dbl>
##  1 rmse    standard       0.135
##  2 rmse    standard       0.137
##  3 rmse    standard       0.133
##  4 rmse    standard       0.127
##  5 rmse    standard       0.130
##  6 rmse    standard       0.126
##  7 rmse    standard       0.126
##  8 rmse    standard       0.136
##  9 rmse    standard       0.131
## 10 rmse    standard       0.140
## # ... with 15 more rows
mean(results$.estimate)
## [1] 0.13