[FIXED] I can't fit Logistic Regression into model. How to solve this?

Issue

I ran this code in Google Colab:

from sklearn.metrics import confusion_matrix

# Initialize logreg model
logreg = LogisticRegression()

# Fit the model with data
logreg.fit(X_train, y_train)

# Predict model
y_pred = logreg.predict(X_test)

# Evaluate model using confusion matrix
cnf_matrix = confusion_matrix(y_test,y_pred)
print('Confusion Matrix:\n', cnf_matrix)

and it gave me this output

<ipython-input-73-2bdb42bd97ad> in <module>()
      6 
      7 # Fit the model with data
----> 8 logreg.fit(X_train, y_train)
      9 
     10 # Predict model

1 frames
/usr/local/lib/python3.7/dist-packages/sklearn/utils/multiclass.py in check_classification_targets(y)
    167     if y_type not in ['binary', 'multiclass', 'multiclass-multioutput',
    168                       'multilabel-indicator', 'multilabel-sequences']:
--> 169         raise ValueError("Unknown label type: %r" % y_type)
    170 
    171 

ValueError: Unknown label type: 'unknown'

How to solve this?

My X variable are numeric columns and y variable is label column. I've tried this code LogisticRegression().fit(X_train, y_train) but it returned error also.

Solution

Sklearn doesn't recognize Boolean type target variables. Convert them to numerical to train:

y_train = y_train.astype('int')

If you want your predictions to show boolean values (instead of integers), you could later convert them to Boolean:

y_pred = y_pred.astype('bool')

Note: if you do decide to convert your predictions, make sure you're not predicting probabilities, but classes (i.e. output is 0 and 1, not a 2-dimensional matrix of intermediate values; if you do predict probabilities, first convert them to classes, then do boolean conversion).

Answered By - dm2