AI0-001 AI Security, Ethics and Governance Practice Question
A data scientist trains a sentiment analysis model on user reviews. To ensure transparency, they want to explain why the model classified a particular review as negative. Which explainability technique should they use?
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
SHAP (SHapley Additive exPlanations)
SHAP (SHapley Additive exPlanations) provides per-feature attribution for individual predictions, making it suitable for explaining why a particular review was classified as negative. Option A is incorrect because a decision tree surrogate model is a global explanation method, not a local explanation for a single instance. Option B is incorrect because PCA is a dimensionality reduction technique, not an explainability method. Option D is incorrect because t-SNE is used for high-dimensional data visualization, not for explaining model predictions.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Decision tree surrogate model
Why it's wrong here
A decision tree surrogate approximates the model's global logic across the dataset; it does not attribute a single review's negative classification to its specific words. It is tempting because its rules are readable, which suits explaining overall model behaviour, not one local prediction.
- ✗
Principal component analysis
Why it's wrong here
Principal component analysis reduces feature dimensionality for visualisation or preprocessing; it produces no per-prediction attribution linking words in a review to the negative class. It is tempting because it exposes which components drive variance across the dataset, which suits exploratory analysis, not explaining an individual classification.
- ✓
SHAP (SHapley Additive exPlanations)
Why this is correct
SHAP assigns each input feature a Shapley value quantifying its contribution to a specific prediction, producing local, per-instance explanations. For a single review classified negative, this pinpoints which words drove the outcome, satisfying the transparency requirement better than global or inherently interpretable alternatives.
- ✗
t-SNE dimensionality reduction
Why it's wrong here
t-SNE projects high-dimensional data into two or three dimensions for cluster visualisation, giving no per-instance feature attribution for a single review's negative label. It is tempting because it reveals how reviews group in embedding space, which supports dataset exploration rather than explaining one prediction.
About these practice questions
One of 962 original AI0-001 practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI0-001 practice question is part of Courseiva's free CompTIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI0-001 exam.