Value error: too many values to pack - in a decision tree algorithm

Question

I've copied the code that is used for data visualization in kaggle. I applied it with another dataset. When I was executing for confusion matrix, visualization etc. it shows value error: too many values to pack (expected 4). I've searched many websites and videos for this error, it explains only for simple python problems not for a visualization. I don't know what all the values are to be added and removed in this code to solve this error.

from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
import collections

from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix, classification_report, roc_auc_score, roc_curve, accuracy_score
conf_matrix_all = {}
a = []
def prediction(name, algo, training_x, testing_x, training_y, testing_y, plot) :
    global a
    algo.fit(training_x,training_y)                           # Fit the training data set to the algorithm passed.
    predictions = algo.predict(testing_x)                     # Get all predictions
    probabilities = algo.predict_proba(testing_x)             # Get probablities of predictions

    conf_matrix = confusion_matrix(testing_y, predictions)    # Get confusion matrix using the predictions
    tn, fp, fn, tp = conf_matrix.ravel()
    
    conf_matrix_all[name] = conf_matrix                       # Save confusion matrix values to a dictionary
    a = conf_matrix    
    
    print("Classification report:")                           # Print the classification report
    print(classification_report(testing_y, predictions))
  
    model_roc_auc = roc_auc_score(testing_y, predictions)           # Get the Area under the curve number
    fpr,tpr,thresholds = roc_curve(testing_y, probabilities[:,1])   # Get False postive rate and true positive rate

    print ("Area under the curve: ", model_roc_auc)
    print(accuracy_score(testing_y, predictions))
    
    if plot:
        fig, axes = plt.subplots(1,2, figsize=(25, 5))
        conf_matrix = np.flip(conf_matrix)
        
        conf_2 = conf_matrix.astype(str)
        labels = np.array([['\nTP','\nFN'],['\nFP','\nTN']])
        labels = np.core.defchararray.add(conf_2, labels)
        sns.heatmap(conf_matrix, fmt='', annot = labels, ax=axes[0], cmap="YlGnBu", xticklabels=[1, 0], yticklabels=[1, 0]);                                           # Plot the confusion matrix
        axes[0].set(xlabel='Predicted', ylabel='Actual')

        plt.title('Receiver Operating Characteristic')
        sns.lineplot(fpr, tpr, ax=axes[1])                                         # Plot the ROC curve
        plt.plot([0, 1], [0, 1],'--')                                              # Plot the diagonal line
        axes[1].set_xlim([0, 1])                                                   # Set x-axis limit to 0 and 1
        axes[1].set_ylim([0, 1])                                                   # Set y-axis limit to 0 and 1
        axes[1].set(xlabel = 'False Positive Rate', ylabel = 'True Positive Rate');
        plt.show();



dtc = DecisionTreeClassifier(criterion='gini', splitter='best', max_depth=10, min_samples_split=2, 
                             min_samples_leaf=1, min_weight_fraction_leaf=0.0, max_features=None, 
                             random_state=None, max_leaf_nodes=None, min_impurity_decrease=0,class_weight=None, ccp_alpha=0.0)

prediction("Decision Tree", dtc, train_X, test_X, train_y, test_y, plot=True)

---

ValueError                                Traceback (most recent call last)
<ipython-input-75-79b3eb994e92> in <module>
      3                              random_state=None, max_leaf_nodes=None, min_impurity_decrease=0,class_weight=None, ccp_alpha=0.0)
      4 
----> 5 prediction("Decision Tree", dtc, train_X, test_X, train_y, test_y, plot=True)

<ipython-input-74-590eb3298a78> in prediction(name, algo, training_x, testing_x, training_y, testing_y, plot)
     14 
     15     conf_matrix = confusion_matrix(testing_y, predictions)    # Get confusion matrix using the predictions
---> 16     tn, fp, fn, tp = conf_matrix.ravel()
     17 
     18     conf_matrix_all[name] = conf_matrix                       # Save confusion matrix values to a dictionary

ValueError: too many values to unpack (expected 4)

I've tried to add.items() or.itervalues() as it mentioned in the videos and websites. i can't figure where it need to be attached.

I want output like this, the detailed classification report. Classification report:

          precision    recall  f1-score   support

       0       0.91      0.71      0.79        41
       1       0.76      0.93      0.84        41

accuracy                           0.82        82
macro avg      0.83      0.82      0.81        82
weighted avg   0.83      0.82      0.81        82

Area under the curve:  0.8170731707317073

Answer 1

I see you labeled the question with multiclass-classification . However, scikit-learns's confusion_matrix() returns an n_classes * n_classes matrix.

You can't assign 9 or 16 or however many values you have here to tn, fp, fn, tp . That terminology is for binary classification.

You can either calculate the metrics you want from the raw multiclass confusion matrix, or use some of the other methods in sklearn.metrics

Answer 2

Check the dimensions of the confusion matrix to ensure that it is a 2x2 matrix, and the dimensions of the predicted and actual labels to ensure that they match.