如何构建混淆矩阵？KNN交叉验证代码混淆矩阵报错求解

Let's break down your issues and fix them step by step:

1. Why You're Getting the "Too Many Values to Unpack" Error #

The root problem here is a misunderstanding of what confusion_matrix.ravel() returns. Since you have 9 target classes, your confusion matrix is a 9x9 grid (81 total elements). When you call .ravel(), it flattens this grid into a 1D array of 81 values—but you're trying to unpack all 81 into just 9 variables. That's why you see ValueError: too many values to unpack (expected 9).

2. Correct Way to Build and Use the Confusion Matrix #

First, let's ensure you compute the confusion matrix correctly, with explicit class labels to avoid ordering issues. Then, we'll handle the results properly:

Step 1: Compute the Confusion Matrix #

Define your class labels explicitly to guarantee the matrix uses the order you expect:

from sklearn.metrics import confusion_matrix
import numpy as np

# List your 9 classes in the order you want them displayed
class_labels = [11, 12, 13, 21, 22, 23, 31, 32, 33]

# Compute the confusion matrix with specified labels
cm = confusion_matrix(y_test, preds, labels=class_labels)

Step 2: Extract Useful Metrics (Instead of Unpacking) #

If you want to get metrics like true positives for each class, use the diagonal of the matrix:

# True positives for each class (diagonal elements)
true_positives = np.diag(cm)
print("True Positives per Class:")
for cls, tp in zip(class_labels, true_positives):
    print(f"Class {cls}: {tp}")

If you need the full flattened matrix, assign it to a single variable:

flattened_cm = cm.ravel()
# Access individual elements via index, e.g., flattened_cm[0] for the first element

Step 3: Print a Readable Confusion Matrix #

For better readability, print the matrix with class labels:

print("\nConfusion Matrix:")
print("Row = True Class, Column = Predicted Class")
print(f"Labels: {class_labels}")
print(cm)

3. Bonus: Visualize the Confusion Matrix #

For a clearer view of your model's performance across all 9 classes, use a heatmap:

import seaborn as sns
import matplotlib.pyplot as plt

plt.figure(figsize=(10, 8))
sns.heatmap(cm, annot=True, fmt='d', cmap='Blues',
            xticklabels=class_labels, yticklabels=class_labels)
plt.xlabel('Predicted Class')
plt.ylabel('True Class')
plt.title('Confusion Matrix for 9-Class KNN Model')
plt.show()

4. Full Fixed Code Snippet #

Here's how the relevant part of your code should look after fixes:

# ... (your existing code up to preds = clf.best_estimator_.predict(X_test))

from sklearn.metrics import confusion_matrix
import numpy as np

class_labels = [11, 12, 13, 21, 22, 23, 31, 32, 33]
cm = confusion_matrix(y_test, preds, labels=class_labels)

# Get true positives per class
true_positives = np.diag(cm)
print("True Positives:", dict(zip(class_labels, true_positives)))

# Print full confusion matrix
print("\nConfusion Matrix:")
print(cm)

Content of the question来源于stack exchange，提问作者Codrin Mironiuc