Open
Description
Describe the bug
I fit a decision forest with training data that includes feature names. When I call predict_proba on the forest everything is fine. When I call rf.estimators_[0].predict_proba it will warn that it was not trained with feature names.
Steps/Code to Reproduce
import numpy as np
from sklearn.ensemble import RandomForestClassifier
print(sklearn.__version__)
data = np.random.normal([1,2,3,4,5,6,7,8,9,10], size=(1000, 10))
feature_names = [f'F{i}' for i in range(10)]
df = pd.DataFrame(data=data, columns=feature_names)
y = np.ones(1000)
y[500:] = 0
rf = RandomForestClassifier()
rf.fit(df, y)
print(f"Feature names in Forest: {rf.feature_names_in_} ")
print('Classing pred proba on forest')
rf.predict_proba(df)
print('Done calling pred proba on forest')
# THIS GIVES A WARNING
rf.estimators_[0].predict_proba(df)Expected Results
No warnings
Actual Results
Feature names in Forest: ['F0' 'F1' 'F2' 'F3' 'F4' 'F5' 'F6' 'F7' 'F8' 'F9']
Classing pred proba on forest
Done calling pred proba on forest
.....lib/python3.8/site-packages/sklearn/base.py:432: UserWarning: X has feature names, but DecisionTreeClassifier was fitted without feature names
warnings.warn(
Versions
sk>>> sklearn.show_versions()
System:
python: 3.8.16 (default, Mar 2 2023, 03:21:46) [GCC 11.2.0]
executable: ....../bin/python
machine: Linux-5.15.0-60-generic-x86_64-with-glibc2.17
Python dependencies:
sklearn: 1.2.2
pip: 23.0.1
setuptools: 65.6.3
numpy: 1.23.1
scipy: 1.9.0
Cython: None
pandas: 1.5.3
matplotlib: None
joblib: 1.1.1
threadpoolctl: 2.2.0Metadata
Metadata
Assignees
Labels
Anything that requires some knowledge of conventions and best practicesAnything that requires some knowledge of conventions and best practices