Logistische regressiemodellen

Machine Learning in de tidyverse

Dmitriy (Dima) Gorenshteyn

Lead Data Scientist, Memorial Sloan Kettering Cancer Center

Binaire classificatie

De attrition-dataset

Logistische regressie

glm(formula = ___, data = ___, family = "binomial")

glm()

head(cv_data)

# A tibble: 5 x 4
  splits       id    train                   validate               
* <list>       <chr> <list>                  <list>                 
1 <S3: rsplit> Fold1 <data.frame [882 × 31]> <data.frame [221 × 31]>
2 <S3: rsplit> Fold2 <data.frame [882 × 31]> <data.frame [221 × 31]>
3 <S3: rsplit> Fold3 <data.frame [882 × 31]> <data.frame [221 × 31]>
4 <S3: rsplit> Fold4 <data.frame [883 × 31]> <data.frame [220 × 31]>
5 <S3: rsplit> Fold5 <data.frame [883 × 31]> <data.frame [220 × 31]>

cv_models_lr <- cv_data %>% 
  mutate(model = map(train, ~glm(formula = Attrition~., 
                                 data = .x, family = "binomial")))

Laten we oefenen!

Machine Learning in de tidyverse