[Python] HistGradientBoostingClassifier Using Categorical Variables

I landed on the HistGradientBoostingClassifier because, according to the docs, it has native support for categorical variables without the need for one-hot encoding or label encoding. I tried to run categorical variables through the model but when I fit the model I got the error:

ValueError: could not convert string to float: 'A'


here is a basic code example of what I am trying to do:

X = pd.DataFrame({
'feature_1': [1, 2, 3, 1, 2],
'feature_2': ['A', 'B', 'C', 'A', 'B']
})

y = pd.Series([0, 1, 0, 1, 0])

X['feature_2'] = X['feature_2'].astype('category')

categorical_features = [1]

model = HistGradientBoostingClassifier(categorical_features=categorical_features)

model.fit(X, y)


I do not have ordinal categorical variables, so I am hesitant to do label encoding. Is there something I am missing here?

Continue reading...