def _fit_strategy_validator(self, X):
"""Internal helper method to validate strategies appropriate for fit.
Checks whether strategies match with type of column they are applied
to. If not, error is raised through `check_strategy_fit` method.
"""
# remove nan columns and store colnames
cols = X.columns.tolist()
self._strats = check_strategy_fit(self.strategy, cols)
# if predictors is a list...
if isinstance(self.predictors, (tuple, list)):
# and it is not the same list of predictors for every iteration...
if not all([isinstance(x, str) for x in self.predictors]):
len_pred = len(self.predictors)
# raise error if not the correct length
if len_pred != self.n:
err = f"Predictors has {len_pred} items. Need {self.n}"
raise ValueError(err)
# check predictors for each in list
self._preds = [
check_predictors_fit(p, cols)
for p in self.predictors
]
# if it is a list, but not a list of objects...
else:
# broadcast predictors
self._preds = check_predictors_fit(self.predictors, cols)
self._preds = [self._preds]*self.n
# if string or dictionary...
else:
# broadcast predictors
self._preds = check_predictors_fit(self.predictors, cols)
self._preds = [self._preds]*self.n
def _fit_strategy_validator(self, X):
"""Private method to validate strategies appropriate for fit.
Checks whether strategies match with type of column they are applied
to. If not, error is raised through `check_strategy_fit` method.
"""
# remove nan columns and store colnames
cols = X.columns.tolist()
self._strats = check_strategy_fit(self.strategy, cols)
self._preds = check_predictors_fit(self.predictors, cols)
def _fit_strategy_validator(self, X):
"""Internal helper method to validate behavior appropriate for fit."""
# remove nan columns and store colnames
cols = X.columns.tolist()
self._preds = check_predictors_fit(self.predictors, cols)
# next, prep the categorical / numerical split
# only necessary for classes that use other features
# wont see this requirement in the single imputer
self.data_mi = X.isnull().astype(int)