tidymodels package and the palmerpenguins datasettidymodels workflow, recipe and modelFor this codelab, you will need to install the tidymodels and palmerpenguins packages.
The tidymodels meta-package is actually a collection of packages for machine learning, that are designed to work together well.
The palmerpenguins package contains a dataset of observed penguin species and their various physical characteristics. The aim of this codelab is to build a model to predict the penguin;s species based on these recorded physical characteristics.
To install the required packages, using the following code in the R console.
install.packages(c("tidymodels", "palmerpenguins"))
Hello
library(tidymodels, quietly = T)
## Registered S3 method overwritten by 'tune':
## method from
## required_pkgs.model_spec parsnip
## ── Attaching packages ────────────────────────────────────── tidymodels 0.1.3 ──
## ✓ broom 0.7.8 ✓ recipes 0.1.16
## ✓ dials 0.0.9 ✓ rsample 0.1.0
## ✓ dplyr 1.0.7 ✓ tibble 3.1.2
## ✓ ggplot2 3.3.5 ✓ tidyr 1.1.3
## ✓ infer 0.5.4 ✓ tune 0.1.5
## ✓ modeldata 0.1.0 ✓ workflows 0.2.2
## ✓ parsnip 0.1.6 ✓ workflowsets 0.0.2
## ✓ purrr 0.3.4 ✓ yardstick 0.0.8
## ── Conflicts ───────────────────────────────────────── tidymodels_conflicts() ──
## x purrr::discard() masks scales::discard()
## x dplyr::filter() masks stats::filter()
## x dplyr::lag() masks stats::lag()
## x recipes::step() masks stats::step()
## • Use tidymodels_prefer() to resolve common conflicts.
library(palmerpenguins, quietly = T)
df_split <- initial_split(penguins, strata = species)
print(df_split)
## <Analysis/Assess/Total>
## <258/86/344>
df_analysis <- df_split %>% analysis()
df_assess <- df_split %>% assessment()
penguins %>%
summarise(across(everything(), ~sum(is.na(.x))))
## # A tibble: 1 x 8
## species island bill_length_mm bill_depth_mm flipper_length_... body_mass_g sex
## <int> <int> <int> <int> <int> <int> <int>
## 1 0 0 2 2 2 2 11
## # ... with 1 more variable: year <int>
rcp <-
recipe(species ~ ., data = df_analysis)
rcp <-
recipe(species ~ ., data = df_analysis) %>%
step_impute_mean(all_numeric_predictors()) %>%
step_impute_mode(all_nominal_predictors())
rcp %>% prep() %>% bake(df_analysis)
## # A tibble: 258 x 8
## island bill_length_mm bill_depth_mm flipper_length_... body_mass_g sex year
## <fct> <dbl> <dbl> <int> <int> <fct> <int>
## 1 Torger... 39.1 18.7 181 3750 male 2007
## 2 Torger... 40.3 18 195 3250 fema... 2007
## 3 Torger... 44.0 17.2 201 4212 male 2007
## 4 Torger... 39.3 20.6 190 3650 male 2007
## 5 Torger... 38.9 17.8 181 3625 fema... 2007
## 6 Torger... 39.2 19.6 195 4675 male 2007
## 7 Torger... 34.1 18.1 193 3475 male 2007
## 8 Torger... 42 20.2 190 4250 male 2007
## 9 Torger... 37.8 17.1 186 3300 male 2007
## 10 Torger... 37.8 17.3 180 3700 male 2007
## # ... with 248 more rows, and 1 more variable: species <fct>
classifier <-
decision_tree(mode = "classification") %>%
set_engine("rpart")
wf <-
workflow() %>%
add_recipe(rcp) %>%
add_model(classifier)
model <- fit(wf, data = df_analysis)
model
## ══ Workflow [trained] ══════════════════════════════════════════════════════════
## Preprocessor: Recipe
## Model: decision_tree()
##
## ── Preprocessor ────────────────────────────────────────────────────────────────
## 2 Recipe Steps
##
## • step_impute_mean()
## • step_impute_mode()
##
## ── Model ───────────────────────────────────────────────────────────────────────
## n= 258
##
## node), split, n, loss, yval, (yprob)
## * denotes terminal node
##
## 1) root 258 144 Adelie (0.441860465 0.197674419 0.360465116)
## 2) flipper_length_mm< 207.5 163 49 Adelie (0.699386503 0.294478528 0.006134969)
## 4) bill_length_mm< 44.65 116 5 Adelie (0.956896552 0.043103448 0.000000000) *
## 5) bill_length_mm>=44.65 47 4 Chinstrap (0.063829787 0.914893617 0.021276596) *
## 3) flipper_length_mm>=207.5 95 3 Gentoo (0.000000000 0.031578947 0.968421053) *
df_pred <-
bind_cols(
df_assess,
predict(model, new_data = df_assess)
)
df_pred %>%
accuracy(truth = species, estimate = .pred_class)
## # A tibble: 1 x 3
## .metric .estimator .estimate
## <chr> <chr> <dbl>
## 1 accuracy multiclass 0.930
df_pred %>%
conf_mat(truth = species, estimate = .pred_class)
## Truth
## Prediction Adelie Chinstrap Gentoo
## Adelie 36 1 1
## Chinstrap 0 15 1
## Gentoo 2 1 29
rcp <-
recipe(species ~ ., data = df_analysis) %>%
step_impute_mean(all_numeric_predictors()) %>%
step_impute_mode(all_nominal_predictors()) %>%
themis::step_upsample(species)
## Registered S3 methods overwritten by 'themis':
## method from
## bake.step_downsample recipes
## bake.step_upsample recipes
## prep.step_downsample recipes
## prep.step_upsample recipes
## tidy.step_downsample recipes
## tidy.step_upsample recipes
## tunable.step_downsample recipes
## tunable.step_upsample recipes
classifier <- rand_forest(mode = "classification")
wf <-
workflow() %>%
add_recipe(rcp) %>%
add_model(classifier)
model <- fit(wf, data = df_analysis)
## Warning: Engine set to `ranger`.
df_pred <-
bind_cols(
df_assess,
predict(model, new_data = df_assess)
)
df_pred %>%
accuracy(truth = species, estimate = .pred_class)
## # A tibble: 1 x 3
## .metric .estimator .estimate
## <chr> <chr> <dbl>
## 1 accuracy multiclass 0.988
df_pred %>%
conf_mat(truth = species, estimate = .pred_class)
## Truth
## Prediction Adelie Chinstrap Gentoo
## Adelie 38 0 1
## Chinstrap 0 17 0
## Gentoo 0 0 30