Amazon.com - Employee Access Challenge--Kaggle

This workflow demonstrates use of Auto Categorical Features Embedding component. The component transforms categorical features into numeric features through three metods: count-encoding, rank of count-encodings and target-encodings.
Each one of the three methods transforms a column of categorical feature to one column of numeric values. This is unlike one-hot-encoding where each categorical column gets transformed to multiple numeric columns.
Dataset is imbalanced and has 10 columns. All columns are categorical. And some columns have very large number of levels.
F1 score for minority class is better when modlling using this compoenet than without it.

External resources

• Data Source--Kaggle