WGU C951 TASK 1: MACHINE
LEARNING PROJECT PROPOSAL FOR
EMAIL SPAM FILTERING
1. In the context of a machine learning project proposal for spam filtering, which type of
learning is most appropriate given a dataset of labeled ‘spam’ and ‘ham’ emails?
A. Unsupervised Learning
B. Semi-supervised Learning
C. Reinforcement Learning
D. Supervised Learning
Answer: D
Conceptual Explanation: Supervised learning is appropriate because the model is trained
on a labeled dataset where the target output (spam or ham) is already known.
2. Which evaluation metric is critical when the cost of a ‘False Positive’ (classifying a
legitimate email as spam) is extremely high for the user?
A. Recall
B. Precision
,C. Accuracy
D. L2 Norm
Answer: B
Conceptual Explanation: Precision measures the ratio of true positives to the total
predicted positives. High precision ensures that very few legitimate emails are incorrectly
sent to the spam folder.
3. What is the primary purpose of ‘Stop Word Removal’ in the data preprocessing phase of a
spam detection proposal?
A. To encrypt the dataset for privacy reasons
B. To identify the root form of every word
C. To remove words that carry little semantic value and reduce noise
D. To increase the dimensionality of the feature space
Answer: C
Conceptual Explanation: Stop words like ‘the’, ‘is’, and ‘at’ are frequent but usually carry
no significant meaning for distinguishing between spam and ham, so removing them
reduces noise.
4. When proposing the use of TF-IDF (Term Frequency-Inverse Document Frequency), what
does the ‘IDF’ component specifically help to address?
A. The penalization of common words that appear in many documents
, B. The importance of words that appear frequently across all documents
C. The frequency of a word within a single document
D. The conversion of categorical labels into numerical values
Answer: A
Conceptual Explanation: IDF reduces the weight of words that occur very frequently
across the entire corpus, highlighting terms that are more unique to specific documents.
5. Which algorithm is often used as a baseline for text classification due to its assumption of
independence between features?
A. Support Vector Machines
B. Naive Bayes
C. Random Forest
D. K-Nearest Neighbors
Answer: B
Conceptual Explanation: Naive Bayes assumes that the presence of a particular feature in
a class is unrelated to the presence of any other feature, making it computationally efficient
and a strong baseline for NLP.
6. In a spam filtering proposal, what is the consequence of ‘Data Leakage’ during the training
process?
A. The model becomes too simple to learn the patterns
LEARNING PROJECT PROPOSAL FOR
EMAIL SPAM FILTERING
1. In the context of a machine learning project proposal for spam filtering, which type of
learning is most appropriate given a dataset of labeled ‘spam’ and ‘ham’ emails?
A. Unsupervised Learning
B. Semi-supervised Learning
C. Reinforcement Learning
D. Supervised Learning
Answer: D
Conceptual Explanation: Supervised learning is appropriate because the model is trained
on a labeled dataset where the target output (spam or ham) is already known.
2. Which evaluation metric is critical when the cost of a ‘False Positive’ (classifying a
legitimate email as spam) is extremely high for the user?
A. Recall
B. Precision
,C. Accuracy
D. L2 Norm
Answer: B
Conceptual Explanation: Precision measures the ratio of true positives to the total
predicted positives. High precision ensures that very few legitimate emails are incorrectly
sent to the spam folder.
3. What is the primary purpose of ‘Stop Word Removal’ in the data preprocessing phase of a
spam detection proposal?
A. To encrypt the dataset for privacy reasons
B. To identify the root form of every word
C. To remove words that carry little semantic value and reduce noise
D. To increase the dimensionality of the feature space
Answer: C
Conceptual Explanation: Stop words like ‘the’, ‘is’, and ‘at’ are frequent but usually carry
no significant meaning for distinguishing between spam and ham, so removing them
reduces noise.
4. When proposing the use of TF-IDF (Term Frequency-Inverse Document Frequency), what
does the ‘IDF’ component specifically help to address?
A. The penalization of common words that appear in many documents
, B. The importance of words that appear frequently across all documents
C. The frequency of a word within a single document
D. The conversion of categorical labels into numerical values
Answer: A
Conceptual Explanation: IDF reduces the weight of words that occur very frequently
across the entire corpus, highlighting terms that are more unique to specific documents.
5. Which algorithm is often used as a baseline for text classification due to its assumption of
independence between features?
A. Support Vector Machines
B. Naive Bayes
C. Random Forest
D. K-Nearest Neighbors
Answer: B
Conceptual Explanation: Naive Bayes assumes that the presence of a particular feature in
a class is unrelated to the presence of any other feature, making it computationally efficient
and a strong baseline for NLP.
6. In a spam filtering proposal, what is the consequence of ‘Data Leakage’ during the training
process?
A. The model becomes too simple to learn the patterns
