Model Interpretability MCQs | JoinMyTutor
MCQs on interpreting ML/DL models.
Model Interpretability
MCQs on model explainability and interpretation.
-
Model interpretability refers to?
- A. Understanding how models make predictions
- B. Ignoring model internals
- C. Only deploying models
- D. Reducing dataset size
Answer: A — Interpretability allows humans to trust and debug AI models.
-
Techniques include?
- A. SHAP, LIME, feature importance, partial dependence
- B. Only training more epochs
- C. Only hyperparameter tuning
- D. Only data cleaning
Answer: A — These techniques reveal the contribution of inputs to predictions.
-
Why is interpretability important?
- A. To detect bias, improve trust, and meet regulatory requirements
- B. Only for academic interest
- C. To slow deployment
- D. No reason
Answer: A — Transparent models help ensure fairness and compliance.
-
Global interpretability shows?
- A. Model behavior over entire dataset
- B. One sample prediction only
- C. Hyperparameter tuning
- D. Data preprocessing
Answer: A — Global interpretability summarizes overall model patterns.
-
Local interpretability shows?
- A. Impact of features on individual predictions
- B. Model summary only
- C. Training loss curve
- D. None
Answer: A — Local interpretability explains why a model predicted a specific output.
-
What is model interpretability?
- A. Degree to which humans can understand model predictions
- B. Model accuracy
- C. Data preprocessing
- D. Hyperparameter tuning
Answer: A — Interpretability focuses on clarity of decision-making process.
-
Techniques for interpretability?
- A. Feature importance, partial dependence plots, surrogate models
- B. Only retraining
- C. Only batch processing
- D. No techniques
Answer: A — Various visualization and approximation methods help interpret models.
-
Global vs local interpretability?
- A. Global explains overall model behavior, local explains single prediction
- B. Local explains model globally
- C. No difference
- D. Global is for single predictions
Answer: A — Different methods target different levels of understanding.
-
Why is interpretability challenging?
- A. Complex models, high-dimensional data, interactions between features
- B. Because models are linear
- C. Because data is simple
- D. No challenge
Answer: A — Complexity increases the difficulty of human understanding.
-
Applications?
- A. Debugging, fairness assessment, regulatory compliance, trust
- B. Only training
- C. Only evaluation
- D. No application
Answer: A — Interpretability helps ensure responsible AI deployment.
-
What does SHAP stand for?
- A. SHapley Additive exPlanations
- B. Simple Heuristic AI Prediction
- C. Structured Hyperplane Analysis
- D. Stochastic Hidden Algorithm Predictor
Answer: A — SHAP is a game-theoretic approach to explain model predictions.
-
Key idea of SHAP?
- A. Allocate contribution of each feature to a prediction fairly
- B. Ignore feature importance
- C. Randomly assign weights
- D. Only global explanation
Answer: A — SHAP uses Shapley values to quantify feature impact.
-
Global vs local SHAP?
- A. Local explains a single prediction, global summarizes feature importance across dataset
- B. Global explains single prediction
- C. No difference
- D. Local ignores features
Answer: A — SHAP can provide both detailed and summary explanations.
-
Advantages of SHAP?
- A. Consistent, model-agnostic, theoretically sound
- B. Only model-specific
- C. Inconsistent results
- D. No advantage
Answer: A — SHAP provides reliable and interpretable explanations for various models.
-
Common visualizations in SHAP?
- A. Force plots, summary plots, dependence plots
- B. Only line plots
- C. Only pie charts
- D. No visualization
Answer: A — Visual tools help understand feature contributions intuitively.
-
What does LIME stand for?
- A. Local Interpretable Model-agnostic Explanations
- B. Linear Model Estimation
- C. Layered Incremental Model Evaluation
- D. Learning Interpretation for Models
Answer: A — LIME explains predictions locally using interpretable surrogate models.
-
Key idea of LIME?
- A. Approximate complex model locally with simple interpretable model
- B. Train another complex model
- C. Ignore local prediction
- D. Use only global summary
Answer: A — LIME focuses on local behavior around a single instance.
-
Difference between LIME and SHAP?
- A. LIME builds local surrogate models; SHAP allocates feature contributions using Shapley values
- B. LIME is global, SHAP is local
- C. No difference
- D. SHAP ignores features
Answer: A — LIME and SHAP have complementary approaches.
-
Advantages of LIME?
- A. Model-agnostic, interpretable, flexible
- B. Model-specific only
- C. Inaccurate
- D. Complex to use
Answer: A — LIME is widely used for explaining black-box predictions.
-
Limitations of LIME?
- A. Approximation may be inaccurate, sensitive to sampling, local scope only
- B. Perfect explanations
- C. Global scope only
- D. No limitation
Answer: A — LIME provides local insights, not full global understanding.
💡 Tips & Common Mistakes
Remember to review concepts carefully. Fill in custom tips for model-interpretability here.
📅 Last Updated: November 20, 2026
Vinod Kumar
Senior Full-Stack InstructorPython & JavaScript Tutor at JoinMyTutor
Vinod teaches practical Python and JavaScript—ranging from basics to advanced web apps—with a focus on clear mental models, code readability, and interview preparation. He authors both the Python and JavaScript tracks on JoinMyTutor.