README.md

In njtierney/broomstick: Convert Decision Tree Objects into Tidy Data Frames

broomstick

Convert decision tree objects into tidy data frames with broomstick.

The goal of broomstick is to extend the broom package to work with decision trees. It is currently borrowing heavily from the prototype package treezy.

Installation

You can install broomstick from github with:

# install.packages("remotes")
remotes::install_github("njtierney/broomstick")

Examples

rpart

library(rpart)
library(broomstick)

fit_rpart <- rpart(Kyphosis ~ Age + Number + Start, 
                   data = kyphosis)

tidy(fit_rpart)
#> # A tibble: 3 × 2
#>   variable importance
#>   <chr>         <dbl>
#> 1 Start          8.20
#> 2 Age            3.10
#> 3 Number         1.52
augment(fit_rpart)
#> # A tibble: 81 × 6
#>    Kyphosis   Age Number Start .fitted[,"absent"] [,"present"] .resid[,"absent"]
#>    <fct>    <int>  <int> <int>              <dbl>        <dbl>             <dbl>
#>  1 absent      71      3     5              0.421        0.579            -0.579
#>  2 absent     158      3    14              0.857        0.143            -0.143
#>  3 present    128      4     5              0.421        0.579            -1.58 
#>  4 absent       2      5     1              0.421        0.579            -0.579
#>  5 absent       1      4    15              1            0                 0    
#>  6 absent       1      2    16              1            0                 0    
#>  7 absent      61      2    17              1            0                 0    
#>  8 absent      37      3    16              1            0                 0    
#>  9 absent     113      2    16              1            0                 0    
#> 10 present     59      6    12              0.429        0.571            -1.57 
#> # ℹ 71 more rows
#> # ℹ 1 more variable: .resid[2] <dbl>

gbm (Boosted Regression Tree)

library(gbm)
#> Loaded gbm 2.1.8.1
library(MASS)
fit_gbm <- gbm(calories ~., data = UScereal)
#> Distribution not specified, assuming gaussian ...

tidy(fit_gbm)
#> # A tibble: 10 × 2
#>    variable importance
#>    <chr>         <dbl>
#>  1 1             25.4 
#>  2 2             22.2 
#>  3 3             17.5 
#>  4 4             11.3 
#>  5 5              8.36
#>  6 6              8.13
#>  7 7              4.99
#>  8 8              2.13
#>  9 9              0   
#> 10 10             0

random forest

library(randomForest)
#> randomForest 4.7-1.1
#> Type rfNews() to see new features/changes/bug fixes.
ozone_rf <- randomForest(Ozone ~ ., 
                         data = airquality, 
                         importance = TRUE,
                         na.action = na.omit)
tidy(ozone_rf)
#> Warning: This function is deprecated as of broom 0.7.0 and will be removed from
#> a future release. Please see tibble::as_tibble().
#> # A tibble: 5 × 4
#>   term    X.IncMSE IncNodePurity imp_sd
#>   <chr>      <dbl>         <dbl>  <dbl>
#> 1 Solar.R    165.         18373.  10.8 
#> 2 Wind       326.         31790.  17.2 
#> 3 Temp       471.         35042.  17.6 
#> 4 Month      109.         10771.   8.90
#> 5 Day         57.7        15353.   9.13
glance(ozone_rf)
#>   mean_mse  mean_rsq
#> 1 336.4239 0.6934116
augment(ozone_rf)
#> Warning in augment.randomForest.method(x, data, ...): casewise importance
#> measures are not available. Run randomForest(..., localImp = TRUE) for more
#> detailed results.
#> # A tibble: 153 × 8
#>    Ozone Solar.R  Wind  Temp Month   Day .oob_times .fitted
#>    <int>   <int> <dbl> <int> <int> <int>      <int>   <dbl>
#>  1    41     190   7.4    67     5     1        191    40.7
#>  2    36     118   8      72     5     2        177    24.0
#>  3    12     149  12.6    74     5     3        191    27.9
#>  4    18     313  11.5    62     5     4        200    24.3
#>  5    NA      NA  14.3    56     5     5         NA    NA  
#>  6    28      NA  14.9    66     5     6         NA    NA  
#>  7    23     299   8.6    65     5     7        186    28.9
#>  8    19      99  13.8    59     5     8        201    19.6
#>  9     8      19  20.1    61     5     9        178    16.1
#> 10    NA     194   8.6    69     5    10         NA    NA  
#> # ℹ 143 more rows

njtierney/broomstick documentation built on Dec. 12, 2023, 5:08 a.m.