MACHINE LEARNING FOR EMAIL
SPAM FILTERING (WGU C951 TASK 3)
1. In the context of email spam filtering, what does the ‘independence’ assumption in a Naive
Bayes classifier imply?
A. Spam and ham emails are mutually exclusive events in the probability space.
B. The classification of one email is independent of the classification of previous emails.
C. The sender’s identity is independent of the content of the email.
D. The presence of one word in an email does not affect the probability of another word
appearing.
Answer: D
Conceptual Explanation: Naive Bayes assumes that features (words) are conditionally
independent given the class label, meaning the occurrence of one word doesn’t influence
the presence of others.
2. Which metric is most critical when the primary goal is to ensure legitimate emails (ham)
are not accidentally moved to the spam folder?
A. Precision
B. Accuracy
,C. Recall
D. Mean Absolute Error
Answer: A
Conceptual Explanation: Precision measures the proportion of predicted positives that
are actually positive. High precision reduces False Positives (ham marked as spam).
3. What is the primary purpose of applying TF-IDF vectorization in a spam detection pipeline?
A. To increase the weight of words that appear frequently across all documents.
B. To convert images found in emails into numerical feature vectors.
C. To down-weight common words that appear frequently across the entire corpus while
highlighting unique identifying words.
D. To automatically correct spelling errors in the email body.
Answer: C
Conceptual Explanation: TF-IDF (Term Frequency-Inverse Document Frequency) reduces
the importance of common words (like ‘the’, ‘and’) and increases the weight of terms
specific to certain categories.
4. During data preprocessing, what is the process of reducing a word to its base or dictionary
form called?
A. Tokenization
B. Lemmatization
, C. Stemming
D. Regularization
Answer: B
Conceptual Explanation: Lemmatization uses linguistic analysis to return words to their
dictionary base (lemma), whereas stemming often just chops off suffixes.
5. Which of the following is considered a ‘False Positive’ in a spam filter?
A. A legitimate email incorrectly identified as spam.
B. A spam email correctly identified as spam.
C. A spam email incorrectly identified as ham.
D. A legitimate email correctly identified as ham.
Answer: A
Conceptual Explanation: A False Positive occurs when the model predicts a positive result
(spam) for a sample that is actually negative (ham).
6. Why is the F1-Score often preferred over Accuracy in evaluating spam filters?
A. Because F1-Score is easier to calculate manually.
B. Because accuracy does not account for the high imbalance between spam and ham in
real-world datasets.
C. Because accuracy is sensitive to the number of features used.
SPAM FILTERING (WGU C951 TASK 3)
1. In the context of email spam filtering, what does the ‘independence’ assumption in a Naive
Bayes classifier imply?
A. Spam and ham emails are mutually exclusive events in the probability space.
B. The classification of one email is independent of the classification of previous emails.
C. The sender’s identity is independent of the content of the email.
D. The presence of one word in an email does not affect the probability of another word
appearing.
Answer: D
Conceptual Explanation: Naive Bayes assumes that features (words) are conditionally
independent given the class label, meaning the occurrence of one word doesn’t influence
the presence of others.
2. Which metric is most critical when the primary goal is to ensure legitimate emails (ham)
are not accidentally moved to the spam folder?
A. Precision
B. Accuracy
,C. Recall
D. Mean Absolute Error
Answer: A
Conceptual Explanation: Precision measures the proportion of predicted positives that
are actually positive. High precision reduces False Positives (ham marked as spam).
3. What is the primary purpose of applying TF-IDF vectorization in a spam detection pipeline?
A. To increase the weight of words that appear frequently across all documents.
B. To convert images found in emails into numerical feature vectors.
C. To down-weight common words that appear frequently across the entire corpus while
highlighting unique identifying words.
D. To automatically correct spelling errors in the email body.
Answer: C
Conceptual Explanation: TF-IDF (Term Frequency-Inverse Document Frequency) reduces
the importance of common words (like ‘the’, ‘and’) and increases the weight of terms
specific to certain categories.
4. During data preprocessing, what is the process of reducing a word to its base or dictionary
form called?
A. Tokenization
B. Lemmatization
, C. Stemming
D. Regularization
Answer: B
Conceptual Explanation: Lemmatization uses linguistic analysis to return words to their
dictionary base (lemma), whereas stemming often just chops off suffixes.
5. Which of the following is considered a ‘False Positive’ in a spam filter?
A. A legitimate email incorrectly identified as spam.
B. A spam email correctly identified as spam.
C. A spam email incorrectly identified as ham.
D. A legitimate email correctly identified as ham.
Answer: A
Conceptual Explanation: A False Positive occurs when the model predicts a positive result
(spam) for a sample that is actually negative (ham).
6. Why is the F1-Score often preferred over Accuracy in evaluating spam filters?
A. Because F1-Score is easier to calculate manually.
B. Because accuracy does not account for the high imbalance between spam and ham in
real-world datasets.
C. Because accuracy is sensitive to the number of features used.
