Confusion matrix and precision/recall quiz

Question 1

What is a confusion matrix in the context of a binary classification problem?

Accepted Answer

A table comparing predicted labels against actual labels

Answer

A graph of accuracy over training epochs

Answer

A list of model hyperparameters

Answer

A matrix of feature correlations

Question 2

In a binary confusion matrix, what does a True Positive (TP) represent?

Accepted Answer

Actual positive predicted as positive

Answer

Actual negative predicted as positive

Answer

Actual positive predicted as negative

Answer

Actual negative predicted as negative

Question 3

What is a False Positive (FP), and why is it sometimes called a Type I error?

Accepted Answer

Predicted positive but actually negative

Answer

Predicted negative but actually positive

Answer

Predicted positive and actually positive

Answer

Predicted negative and actually negative

Question 4

What is a False Negative (FN), and what real-world risk does it carry in medical screening?

Accepted Answer

A real positive case predicted as negative (a missed case)

Answer

A real negative case predicted as positive

Answer

A correct positive prediction

Answer

A correct negative prediction

Question 5

Write the formula for accuracy in terms of TP, TN, FP, and FN.

Accepted Answer

(TP + TN) / (TP + TN + FP + FN)

Answer

TP / (TP + FP)

Answer

TP / (TP + FN)

Answer

(FP + FN) / (TP + TN)

Question 6

Define precision and give its formula.

Accepted Answer

TP / (TP + FP)

Answer

TP / (TP + FN)

Answer

(TP + TN) / total

Answer

TN / (TN + FP)

Question 7

Define recall (sensitivity) and give its formula.

Accepted Answer

TP / (TP + FN)

Answer

TP / (TP + FP)

Answer

TN / (TN + FN)

Answer

FP / (FP + TN)

Question 8

A classifier has TP=80, FP=20, FN=40, TN=860. Calculate its precision and recall.

Accepted Answer

Precision = 0.80, Recall = 0.667

Answer

Precision = 0.667, Recall = 0.80

Answer

Precision = 0.80, Recall = 0.80

Answer

Precision = 0.50, Recall = 0.667

Question 9

Why can accuracy be misleading on an imbalanced dataset?

Accepted Answer

A trivial majority-class model can score high accuracy while ignoring the minority class

Answer

Accuracy cannot be computed for imbalanced data

Answer

Accuracy always equals recall on imbalanced data

Answer

Imbalanced data makes accuracy negative

Question 10

What is the F1 score and why is it used?

Accepted Answer

The harmonic mean of precision and recall

Answer

The arithmetic mean of precision and recall

Answer

The product of accuracy and recall

Answer

The difference between precision and recall

Question 11

A model has precision = 0.6 and recall = 0.6. What is its F1 score?

Accepted Answer

0.6

Answer

0.36

Answer

1.2

Answer

0.5

Question 12

Why is the F1 score a harmonic mean rather than an arithmetic mean of precision and recall?

Accepted Answer

It punishes large gaps between precision and recall, rewarding balance

Answer

It is easier to compute than the arithmetic mean

Answer

It always gives a higher value than the arithmetic mean

Answer

It ignores recall entirely

Question 13

In a spam filter, which error is usually worse: a False Positive or a False Negative, and why?

Accepted Answer

False Positive, because a real email could be lost

Answer

False Negative, because seeing spam is catastrophic

Answer

Both are equally harmless

Answer

Neither matters for spam filters

Question 14

For a cancer-screening test, should you prioritize precision or recall, and why?

Accepted Answer

Recall, because missing a real case is very dangerous

Answer

Precision, because false alarms must be avoided at all costs

Answer

Accuracy, since the classes are balanced

Answer

Specificity only

Question 15

What is the precision-recall trade-off when adjusting a classification threshold?

Accepted Answer

Raising the threshold usually raises precision and lowers recall

Answer

Raising the threshold raises both precision and recall

Answer

The threshold has no effect on precision or recall

Answer

Lowering the threshold always increases precision

Question 16

What does specificity measure, and what is its formula?

Accepted Answer

TN / (TN + FP)

Answer

TP / (TP + FN)

Answer

TP / (TP + FP)

Answer

TN / (TN + FN)

Question 17

If a model predicts every instance as positive, what are its recall and precision likely to be?

Accepted Answer

Recall = 1.0, precision possibly low

Answer

Recall = 0.0, precision = 1.0

Answer

Both equal 1.0

Answer

Both equal 0.0

Question 18

Compute the confusion matrix and precision using scikit-learn for given true and predicted labels.

Accepted Answer

confusion_matrix and precision_score from sklearn.metrics

Answer

plot_tree from sklearn.tree

Answer

train_test_split from sklearn.model_selection

Answer

StandardScaler from sklearn.preprocessing

Question 19

What is the difference between sensitivity and specificity?

Accepted Answer

Sensitivity is the true positive rate; specificity is the true negative rate

Answer

They are two names for the same quantity

Answer

Sensitivity uses negatives; specificity uses positives

Answer

Specificity equals precision

Question 20

In a confusion matrix produced by scikit-learn, what is the standard order of the four cells?

Accepted Answer

[[TN, FP], [FN, TP]]

Answer

[[TP, FP], [FN, TN]]

Answer

[[TP, FN], [FP, TN]]

Answer

[[FN, TP], [TN, FP]]

Question 21

What does the False Positive Rate (FPR) measure, and how is it computed?

Accepted Answer

FP / (FP + TN)

Answer

FP / (FP + TP)

Answer

FN / (FN + TP)

Answer

TN / (TN + FP)

Question 22

A fraud-detection model flags 200 transactions as fraud; 150 are truly fraud, but 50 frauds went undetected. What are its precision and recall?

Accepted Answer

Precision = 0.75, Recall = 0.75

Answer

Precision = 0.50, Recall = 0.75

Answer

Precision = 0.75, Recall = 0.50

Answer

Precision = 1.00, Recall = 0.75

Question 23

How does a confusion matrix extend to multi-class classification problems?

Accepted Answer

It becomes a K x K matrix with correct counts on the diagonal

Answer

It stays 2x2 regardless of class count

Answer

It becomes a single number

Answer

It cannot be defined for more than two classes

Question 24

What is the difference between macro-averaged and micro-averaged precision in multi-class problems?

Accepted Answer

Macro averages per-class scores equally; micro pools all counts first

Answer

Macro pools counts; micro averages per class

Answer

They always give the same value

Answer

Micro ignores small classes; macro ignores large classes

Question 25

If precision = 0.9 and recall = 0.1, what does this tell you about the model's behavior?

Accepted Answer

It is conservative: rarely predicts positive but is usually right when it does

Answer

It catches almost all positives but with many false alarms

Answer

It performs perfectly

Answer

It predicts everything as positive

Question 26

What does the diagonal of a confusion matrix represent?

Accepted Answer

The correctly classified samples

Answer

The misclassified samples

Answer

The model's hyperparameters

Answer

The class probabilities

Question 27

Why might two models with the same accuracy have very different precision and recall?

Accepted Answer

Because accuracy ignores how errors split between FP and FN

Answer

Because accuracy is computed incorrectly

Answer

Because precision and recall are random

Answer

They cannot have the same accuracy

Question 28

What does an F-beta score with beta > 1 emphasize compared to the standard F1 score?

Accepted Answer

It emphasizes recall more than precision

Answer

It emphasizes precision more than recall

Answer

It ignores both precision and recall

Answer

It always equals F1

Question 29

Generate a full classification report (precision, recall, F1) using scikit-learn.

Accepted Answer

classification_report from sklearn.metrics

Answer

describe() from pandas

Answer

summary() from numpy

Answer

report_metrics from sklearn.tree

Question 30

How are precision and recall related to the ROC curve and the precision-recall curve?

Accepted Answer

PR curves are often better than ROC for imbalanced data

Answer

ROC and PR curves are identical

Answer

PR curves ignore recall

Answer

ROC curves plot precision vs recall

Test yourself