Suggestion to improve GLRM sklearn compability

Hello, I'm trying to use the GLRM LogisticRuleRegression and it seems to be compatible with my own code for training/evaluation with sklearn models. However, it fails when I use functions like GridSearchCV for hyperparameter tuning.

TypeError: Cannot clone object '<aix360.algorithms.rbm.logistic_regression.LogisticRuleRegression object at 0x10f731310>' (type <class 'aix360.algorithms.rbm.logistic_regression.LogisticRuleRegression'>): it does not seem to be a scikit-learn estimator as it does not implement a 'get_params' methods. (py37) Jamess-MacBook-Pro-2:sk

If the class inherits from BaseEstimator and ClassifierMixin from sklearn.base instead of just object, then it will inherit get_params() and this will resolve the issue. I've tested this on my local machine. So the change should be:

class LogisticRuleRegression(object):
to
class LogisticRuleRegression(BaseEstimator, ClassifierMixin):

This can also be applied to LinearRuleRegression (replacing ClassifierMixin with RegressorMixin) and any other similar classes and may resolve other sklearn compatability issues I haven't come across yet (e.g. Pipeline may be affected as well)

Start by locating the LogisticRuleRegression and LinearRuleRegression class definitions and reviewing how they are initialized and used with scikit-learn. Verify compatibility with GridSearchCV and Pipeline, then add or update tests showing that the estimators can be cloned and used for hyperparameter tuning.