Mengurangi fitur model

Rekayasa Fitur di R

Jorge Zazueta

Research Professor and Head of the Modeling Group at the School of Economics, UASLP

Alasan mengurangi jumlah fitur

Menghapus variabel yang tidak relevan atau berinformasi rendah dapat memberi manfaat, antara lain

Mengurangi varians model tanpa banyak menambah bias
Meningkatkan kinerja di data uji
Mempercepat komputasi
Menurunkan kompleksitas model
Meningkatkan interpretabilitas

Menyaring data dengan kepentingan variabel

Melatih model dengan semua fitur

lr_recipe_full <-
  recipe(Loan_Status ~., data = train) %>%
  update_role(Loan_ID, new_role = "ID")

lr_workflow_full <- 
  workflow() %>%
  add_model(lr_model) %>%
  add_recipe(lr_recipe_full)

lr_fit_full <- 
  lr_workflow_full %>%
  fit(data = train)

Memplot vip variabel

lr_fit_full %>%
  extract_fit_parsnip() %>%
  vip(aesthetics = list(fill = "steelblue"))

Kepentingan variabel Bagan kolom kepentingan variabel.

Bangun model tereduksi dengan sintaks formula

Kita dapat menambahkan fitur langsung dengan sintaks formula R dasar.

# Buat recipe
recipe_formula <- 
  recipe(Loan_Status ~ Credit_History + Property_Area + 
           LoanAmount, data = train)

# Gabungkan dengan model
workflow_formula <- # Bundle with model
  workflow() %>% add_model(lr_model) %>%
  add_recipe(recipe_formula)

Bangun model tereduksi dengan membuat vektor fitur

Vektor fitur dapat digunakan untuk memilih fitur sebelum pelatihan.

# Feature vector
features <- c("Credit_History", "Property_Area", "LoanAmount", "Loan_Status") 

# Data latih dan uji
train_features <- train %>% select(all_of(features))
test_features <- test %>% select(all_of(features))

# Buat recipe dan gabungkan dengan model
recipe_features <- recipe(Loan_Status ~., data = train_features)
workflow_features <- workflow() %>% add_model(lr_model) %>%
  add_recipe(recipe_features)

Membuat objek augmented

Objek augmented untuk kedua pendekatan

lr_aug_formula <-
  workflow_formula %>%
  fit(data = train) %>%
  augment(new_data = test)

lr_aug_features <-
  workflow_features %>%
  fit(data = train_features) %>%
  augment(new_data = test_features)

Keduanya memberi hasil yang sama

all_equal(lr_aug_features, 
lr_aug_formula %>%
select(all_of(features),
starts_with(".pred")))

[1] TRUE

Membandingkan model penuh vs. tereduksi

Menggunakan semua fitur

lr_fit_full <- # Fit workflow
  lr_workflow_full %>%
  fit(data = train)
lr_aug_full <- # Augment
  lr_fit_full %>%
  augment(test)
lr_aug_full %>% # Evaluasi
  class_evaluate(truth = Loan_Status, 
                 estimate = .pred_class,
                 .pred_Y)

# A tibble: 2 × 3
  .metric  .estimator .estimate
  <chr>    <chr>          <dbl>
1 accuracy binary         0.842
2 roc_auc  binary         0.744

Menggunakan 3 fitur teratas*

lr_fit_formula <- # Fit workflow
  workflow_formula %>%
  fit(train)
lr_aug_formula <- # Augment
  lr_fit_formula %>%
  augment(new_data = test)
lr_aug_formula %>% # Evaluasi
  class_evaluate(truth = Loan_Status, 
                 estimate = .pred_class,
                 .pred_Y)

# A tibble: 2 × 3
  .metric  .estimator .estimate
  <chr>    <chr>          <dbl>
1 accuracy binary         0.842
2 roc_auc  binary         0.733

Ayo berlatih!

Rekayasa Fitur di R