Classification reports in TensorFlow - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of classification report with imbalanced classes

Given a binary classification with true labels and predicted labels as below, what is the precision for class 1 in the classification report?

TensorFlow

from sklearn.metrics import classification_report

y_true = [0, 0, 0, 1, 1, 1, 1]
y_pred = [0, 0, 1, 1, 0, 1, 1]

report = classification_report(y_true, y_pred, output_dict=True)
precision_class_1 = report['1']['precision']
print(round(precision_class_1, 2))

A 0.60

B 0.67

C 0.80

D 0.75

💡 Hint

Precision is the ratio of true positives to all predicted positives for that class.

Precision for class 1 is calculated as TP / (TP + FP). Here, TP=3 (correctly predicted 1s), FP=1 (predicted 1 but true label 0), so precision = 3/(3+1) = 0.75. The code prints round(0.75, 2) = 0.75 (option D).

❓ Model Choice

intermediate

Choosing the right metric for multi-class classification

You have a multi-class classification problem with 5 classes and want to evaluate your TensorFlow model. Which metric from the classification report best shows the balance between precision and recall for each class?

A Accuracy

B F1-score

C Precision

D Recall

💡 Hint

Think about a metric that combines precision and recall.

F1-score is the harmonic mean of precision and recall, showing a balance between the two. It is especially useful when you want to consider both false positives and false negatives.

❓ Hyperparameter

advanced

Effect of threshold on classification report metrics

In a binary classification TensorFlow model, you adjust the decision threshold from 0.5 to 0.7. How does this change most likely affect the precision and recall for the positive class in the classification report?

A Precision increases, recall decreases

B Precision decreases, recall increases

C Both precision and recall increase

D Both precision and recall decrease

💡 Hint

Raising the threshold makes the model more strict about positive predictions.

Increasing the threshold means fewer positive predictions, so false positives reduce, increasing precision. But some true positives are missed, lowering recall.

🔧 Debug

advanced

Debugging unexpected classification report output

You run classification_report on your TensorFlow model predictions but see all zeros for precision, recall, and f1-score for one class. What is the most likely cause?

TensorFlow

from sklearn.metrics import classification_report

true_labels = [0, 1, 2, 2, 1]
pred_labels = [0, 0, 2, 2, 0]

print(classification_report(true_labels, pred_labels))

A The model never predicted that class, causing zero division in metrics

B The true labels are missing that class, so metrics are zero

C The classification_report function is not compatible with TensorFlow

D The labels are not integers, causing metric calculation errors

💡 Hint

Check if the model predicted any samples for that class.

If the model never predicts a class, precision and recall for that class become zero because there are no true positives or predicted positives, leading to zero metrics.

🧠 Conceptual

expert

Interpreting macro vs weighted averages in classification reports

In a multi-class classification report, what is the key difference between macro average and weighted average for precision, recall, and F1-score?

A Both macro and weighted averages weight metrics by class frequency

B Macro average weights metrics by class support; weighted average averages equally

C Macro average calculates metrics per class and averages equally; weighted average weights by class support

D Macro average only considers the largest class; weighted average considers all classes

💡 Hint

Think about how class size affects the averaging.

Macro average treats all classes equally regardless of size, while weighted average accounts for the number of true instances per class, giving more influence to larger classes.