Describe the bug

In _check_partial_fit_first_call, it requires:

Regressors such as DecisionTreeRegressor don't have the classes_ attribute, yet like classifiers, they could also benefit from a partial_fit function. I believe an is_classifier check should solve the problem.

Steps/Code to Reproduce

Run _check_partial_fit_first_call on a regressor with the partial_fit function implemented.

Expected Results

No error is thrown when classes is not passed into partial_fit for regressors.

Actual Results

ValueError: classes must be passed on the first call to partial_fit.

I ran into this error when testing #18889 and had to pass an empty classes parameter.

Versions

System:
    python: 3.8.5 (default, Sep  4 2020, 02:22:02)  [Clang 10.0.0 ]
executable: /Users/pssf23/miniconda3/envs/ndd/bin/python
   machine: macOS-10.16-x86_64-i386-64bit

Python dependencies:
          pip: 20.3.3
   setuptools: 51.0.0.post20201207
      sklearn: 0.24.dev0
        numpy: 1.21.0
        scipy: 1.7.1
       Cython: 0.29.21
       pandas: 1.2.0
   matplotlib: 3.3.3
       joblib: 1.0.0
threadpoolctl: 2.1.0

Built with OpenMP: False