[BUG] KernelExplainer returns TypeError for certain input

[Nanthini10]

Describe the bug
SHAP returns TypeError: X matrix must have at least a column this is likely because none of the features are deemed important enough, so a 0 column input is passed to LR.

Steps/Code to reproduce bug

from cuml.explainer import KernelExplainer
from cuml.datasets import make_regression
from cuml import Lasso
X, y = make_regression(n_samples=100, n_features=10)
clf = Lasso()
clf.fit(X,y)

exp = KernelExplainer(model=clf.predict, data=X, nsamples=10)
exp.shap_values(X)

Error

/opt/conda/envs/rapids/lib/python3.8/site-packages/sklearn/linear_model/_least_angle.py:615: ConvergenceWarning: Regressors in active set degenerate. Dropping a regressor, after 2 iterations, i.e. alpha=5.722e-07, with an active set of 2 regressors, and the smallest cholesky pivot element being 2.220e-16. Reduce max_iter or increase eps parameters.
  warnings.warn('Regressors in active set degenerate. '
/opt/conda/envs/rapids/lib/python3.8/site-packages/sklearn/linear_model/_least_angle.py:615: ConvergenceWarning: Regressors in active set degenerate. Dropping a regressor, after 3 iterations, i.e. alpha=2.861e-07, with an active set of 3 regressors, and the smallest cholesky pivot element being 2.220e-16. Reduce max_iter or increase eps parameters.
  warnings.warn('Regressors in active set degenerate. '
/opt/conda/envs/rapids/lib/python3.8/site-packages/cuml/internals/api_decorators.py:409: UserWarning: Changing solver from 'eig' to 'svd' as eig solver does not support training data with 1 column currently.
  return func(*args, **kwargs)
/opt/conda/envs/rapids/lib/python3.8/site-packages/cuml/internals/api_decorators.py:409: UserWarning: Changing solver from 'eig' to 'svd' as eig solver does not support training data with 1 column currently.
  return func(*args, **kwargs)
/opt/conda/envs/rapids/lib/python3.8/site-packages/sklearn/linear_model/_least_angle.py:642: ConvergenceWarning: Early stopping the lars path, as the residues are small and the current value of alpha is no longer well controlled. 4 iterations, alpha=4.860e-02, previous alpha=4.334e-02, with an active set of 5 regressors.
  warnings.warn('Early stopping the lars path, as the residues '
/opt/conda/envs/rapids/lib/python3.8/site-packages/sklearn/linear_model/_least_angle.py:615: ConvergenceWarning: Regressors in active set degenerate. Dropping a regressor, after 7 iterations, i.e. alpha=1.619e-07, with an active set of 7 regressors, and the smallest cholesky pivot element being 2.220e-16. Reduce max_iter or increase eps parameters.
  warnings.warn('Regressors in active set degenerate. '
/opt/conda/envs/rapids/lib/python3.8/site-packages/sklearn/linear_model/_least_angle.py:615: ConvergenceWarning: Regressors in active set degenerate. Dropping a regressor, after 7 iterations, i.e. alpha=1.245e-07, with an active set of 7 regressors, and the smallest cholesky pivot element being 2.220e-16. Reduce max_iter or increase eps parameters.
  warnings.warn('Regressors in active set degenerate. '

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-1-b6cdbd890780> in <module>
      8 
      9 exp = KernelExplainer(model=clf.predict, data=X, nsamples=10)
---> 10 exp.shap_values(X)

cuml/explainer/kernel_shap.pyx in cuml.explainer.kernel_shap.KernelExplainer.shap_values()

cuml/explainer/base.pyx in cuml.explainer.base.SHAPBase._explain()

cuml/explainer/kernel_shap.pyx in cuml.explainer.kernel_shap.KernelExplainer._explain_single_observation()

cuml/explainer/kernel_shap.pyx in cuml.explainer.kernel_shap._weighted_linear_regression()

/opt/conda/envs/rapids/lib/python3.8/site-packages/cuml/internals/api_decorators.py in inner_with_setters(*args, **kwargs)
    407                                 target_val=target_val)
    408 
--> 409                 return func(*args, **kwargs)
    410 
    411         @wraps(func)

cuml/linear_model/linear_regression.pyx in cuml.linear_model.linear_regression.LinearRegression.fit()

TypeError: X matrix must have at least a column

Expected behavior
It should either select at least one column or return a more descriptive error for the user. @dantegd What are you thoughts?

Environment details (please complete the following information):
docker 21.10 nightly