Q14: not fitted error

Module 4. Linear Models Wrap-up quiz
#1

I’m using this exact snippet from Q14 but get an error. I don’t know why, especially that the first line says fit. I tried to get rid of target parameter - same thing happens. I also copied suggested answer from the previous exercises to make sure that every thing is all right. Any clue?

preprocessor.fit(data, target)
feature_names = (preprocessor.named_transformers_["onehotencoder"]
                             .get_feature_names(categorical_columns)).tolist()
feature_names += numerical_columns
feature_names

---------------------------------------------------------------------------
NotFittedError                            Traceback (most recent call last)
<ipython-input-11-4a32008c25d3> in <module>
----> 1 feature_names = (preprocessor.named_transformers_["onehotencoder"]
      2                              .get_feature_names(categorical_columns)).tolist()
      3 feature_names += numerical_columns
      4 feature_names

/opt/conda/lib/python3.9/site-packages/sklearn/preprocessing/_encoders.py in get_feature_names(self, input_features)
    626             Array of feature names.
    627         """
--> 628         check_is_fitted(self)
    629         cats = self.categories_
    630         if input_features is None:

/opt/conda/lib/python3.9/site-packages/sklearn/utils/validation.py in inner_f(*args, **kwargs)
     61             extra_args = len(args) - len(all_args)
     62             if extra_args <= 0:
---> 63                 return f(*args, **kwargs)
     64 
     65             # extra_args > 0

/opt/conda/lib/python3.9/site-packages/sklearn/utils/validation.py in check_is_fitted(estimator, attributes, msg, all_or_any)
   1096 
   1097     if not attrs:
-> 1098         raise NotFittedError(msg % {'name': type(estimator).__name__})
   1099 
   1100 

NotFittedError: This OneHotEncoder instance is not fitted yet. Call 'fit' with appropriate arguments before using this estimator.
#2

Can you check again if preprocessor.fit(data, target) has been called. The error obtained would be raised if this is not the case.

#3

It is called, by the way in the description of Q14, target is not passed

#4

True the target is not required.

#5

Using the answers from the question preceding the question 14 allows to run the code.
I just tried it.

#6

Could you provide the entire snippet of code with the definition of the preprocessor up to the fit

#7 
from sklearn.compose import make_column_selector as selector
from sklearn.compose import make_column_transformer
from sklearn.preprocessing import OneHotEncoder
from sklearn.model_selection import cross_validate
from sklearn.preprocessing import StandardScaler
import pandas as pd
from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import make_pipeline

adult_census = pd.read_csv("../datasets/adult-census.csv")
target = adult_census["class"]
data = adult_census.select_dtypes(["integer", "floating"])
data = data.drop(columns=["education-num"])

categorical_columns = selector(dtype_include=object)(data)
numerical_columns = selector(dtype_exclude=object)(data)

preprocessor = make_column_transformer(
    (OneHotEncoder(handle_unknown="ignore"), categorical_columns),
    (StandardScaler(), numerical_columns),
)
model = make_pipeline(preprocessor, LogisticRegression(max_iter=500))
cv_results = cross_validate(
    model, data, target, cv=10, return_estimator=True, n_jobs=2
)
cv_results["test_score"].mean()
#8

Hi,
the cause of the problem is you using a different way to prepare the data set.
Before Question 12 in the wrap-up quiz there is this cell:

Now, we will work with both numerical and categorical features. You can load Adult Census with the following snippet:

adult_census = pd.read_csv("../datasets/adult-census.csv")
target = adult_census["class"]
data = adult_census.drop(columns=["class", "education-num"])

If you replace your 4 lines of data loading code with this snippet all is working fine.

#9

Yes, this is the reason. No categorical columns were dispatched to the OneHotEncoder and it results in an unfitted OneHotEncoder. Scikit-learn should be nicer by providing a warning when calling preprocessor.fit(data) stating that one of the transformers did not receive any data to be processed.

#10

Okay, thank you. I see it now :slight_smile: