Adding enable_categorical to .apply method

[imeyer2]

Currently, XGBModel can be trained with enable_categorical=True when the input data fields have categorical type when appropriate. However, the .apply method will throw when trying to retrieve leaf values.

Example

import xgboost as xgb
import numpy as np
import pandas as pd

model = xgb.XGBClassifier(enable_categorical = True)

arr = np.random.rand(5,5)

df = pd.DataFrame(arr)

df['test']= ['one', 'two','three', 'four', 'five']


df = df.convert_dtypes()

# convert string to categorical
df['test'] = df['test'].astype('category')

model.fit(df, [0,1,2,3,4]) # works


model.apply(df) # fails

Produces

Exception has occurred: ValueError
DataFrame.dtypes for data must be int, float, bool or category. When categorical type is supplied, the experimental DMatrix parameter`enable_categorical` must be set to `True`.  Invalid columns:test: category
KeyError: 'category'

During handling of the above exception, another exception occurred:

  File "/path/to/python/file/test.py", line 22, in <module>
    model.apply(df)
ValueError: DataFrame.dtypes for data must be int, float, bool or category. When categorical type is supplied, the experimental DMatrix parameter`enable_categorical` must be set to `True`.  Invalid columns:test: category

[trivialfis]

Thank you for the fix! Could you please help add a small test in the test_with_sklearn.py file?

[imeyer2]

Thank you for the fix! Could you please help add a small test in the test_with_sklearn.py file?

Done!