AI Feature Engineering MCQ Questions with Answers – Page 2 (Latest 2026)

Q51. Which statement about binning/discretization is most accurate?

A. Pure normalization. B. Z-score standardization. C. Polynomial features. D. Convert continuous to categorical bins.

Select an answer to check.

Answer: Convert continuous to categorical bins.

Here, Convert continuous to categorical bins. is the right choice. Helpful for tree models sometimes. It aligns directly with what the question asks about which statement about binning/discretization is most accurate. Competing choices sound plausible, but they miss the key condition.

Q52. How is binning/discretization best characterized?

A. Z-score standardization. B. Polynomial features. C. Pure normalization. D. Convert continuous to categorical bins.

Select an answer to check.

Answer: Convert continuous to categorical bins.

In this case, Convert continuous to categorical bins. is correct. Helpful for tree models sometimes. It aligns directly with what the question asks about how is binning/discretization best characterized. Competing choices sound plausible, but they miss the key condition.

Q53. Which option best describes missing value imputation?

A. Fill missing values with mean/median/mode/model. B. Replace with NaN string. C. Encrypt missing values. D. Always drop rows with missing.

Select an answer to check.

Answer: Fill missing values with mean/median/mode/model.

The best option here is Fill missing values with mean/median/mode/model.. Many strategies; choose carefully. It aligns directly with what the question asks about which option best describes missing value imputation. Competing choices sound plausible, but they miss the key condition.

Q54. What is the primary purpose of missing value imputation?

A. Fill missing values with mean/median/mode/model. B. Encrypt missing values. C. Always drop rows with missing. D. Replace with NaN string.

Select an answer to check.

Answer: Fill missing values with mean/median/mode/model.

For this question, Fill missing values with mean/median/mode/model. is correct. Many strategies; choose carefully. It aligns directly with what the question asks about what is the primary purpose of missing value. Competing choices sound plausible, but they miss the key condition.

Q55. Which statement about missing value imputation is most accurate?

A. Fill missing values with mean/median/mode/model. B. Replace with NaN string. C. Encrypt missing values. D. Always drop rows with missing.

Select an answer to check.

Answer: Fill missing values with mean/median/mode/model.

Fill missing values with mean/median/mode/model. is the correct answer here. Many strategies; choose carefully. It aligns directly with what the question asks about which statement about missing value imputation is most. Competing choices sound plausible, but they miss the key condition.

Q56. How is missing value imputation best characterized?

A. Replace with NaN string. B. Fill missing values with mean/median/mode/model. C. Always drop rows with missing. D. Encrypt missing values.

Select an answer to check.

Answer: Fill missing values with mean/median/mode/model.

Here, Fill missing values with mean/median/mode/model. is the right choice. Many strategies; choose carefully. This matches the core idea being tested around how is missing value imputation best characterized. Competing choices sound plausible, but they miss the key condition.

Q57. Which option best describes indicator-for-missing?

A. Encode as zero only. B. Encode as -1 only. C. Add a binary feature flagging the missing. D. Drop the column.

Select an answer to check.

Answer: Add a binary feature flagging the missing.

In this case, Add a binary feature flagging the missing. is correct. Preserves information about missingness. This matches the core idea being tested around which option best describes indicator-for-missing. Competing choices sound plausible, but they miss the key condition.

Q58. What is the primary purpose of indicator-for-missing?

A. Encode as zero only. B. Add a binary feature flagging the missing. C. Drop the column. D. Encode as -1 only.

Select an answer to check.

Answer: Add a binary feature flagging the missing.

The best option here is Add a binary feature flagging the missing.. Preserves information about missingness. This matches the core idea being tested around what is the primary purpose of indicator-for-missing. Competing choices sound plausible, but they miss the key condition.

Q59. Which statement about indicator-for-missing is most accurate?

A. Add a binary feature flagging the missing. B. Encode as -1 only. C. Drop the column. D. Encode as zero only.

Select an answer to check.

Answer: Add a binary feature flagging the missing.

For this question, Add a binary feature flagging the missing. is correct. Preserves information about missingness. This matches the core idea being tested around which statement about indicator-for-missing is most accurate. Competing choices sound plausible, but they miss the key condition.

Q60. How is indicator-for-missing best characterized?

A. Encode as zero only. B. Add a binary feature flagging the missing. C. Drop the column. D. Encode as -1 only.

Select an answer to check.

Answer: Add a binary feature flagging the missing.

Add a binary feature flagging the missing. is the correct answer here. Preserves information about missingness. This matches the core idea being tested around how is indicator-for-missing best characterized. Competing choices sound plausible, but they miss the key condition.

Q61. Which option best describes feature selection?

A. Choose a subset of features to use. B. Always using all features. C. Random sampling features. D. Hashing all features.

Select an answer to check.

Answer: Choose a subset of features to use.

Here, Choose a subset of features to use. is the right choice. Reduces overfit and cost. That is exactly the concept behind which option best describes feature selection in this context. Competing choices sound plausible, but they miss the key condition.

Q62. What is the primary purpose of feature selection?

A. Always using all features. B. Random sampling features. C. Hashing all features. D. Choose a subset of features to use.

Select an answer to check.

Answer: Choose a subset of features to use.

In this case, Choose a subset of features to use. is correct. Reduces overfit and cost. That is exactly the concept behind what is the primary purpose of feature selection in this context. Competing choices sound plausible, but they miss the key condition.

Q63. Which statement about feature selection is most accurate?

A. Hashing all features. B. Random sampling features. C. Choose a subset of features to use. D. Always using all features.

Select an answer to check.

Answer: Choose a subset of features to use.

The best option here is Choose a subset of features to use.. Reduces overfit and cost. That is exactly the concept behind which statement about feature selection is most accurate in this context. Competing choices sound plausible, but they miss the key condition.

Q64. How is feature selection best characterized?

A. Random sampling features. B. Choose a subset of features to use. C. Hashing all features. D. Always using all features.

Select an answer to check.

Answer: Choose a subset of features to use.

For this question, Choose a subset of features to use. is correct. Reduces overfit and cost. That is exactly the concept behind how is feature selection best characterized in this context. Competing choices sound plausible, but they miss the key condition.

Q65. Which option best describes mutual information?

A. Variance only. B. Pearson correlation only. C. Measures statistical dependence between feature and target. D. Mean only.

Select an answer to check.

Answer: Measures statistical dependence between feature and target.

Measures statistical dependence between feature and target. is the correct answer here. Captures non-linear relationships. That is exactly the concept behind which option best describes mutual information in this context. Competing choices sound plausible, but they miss the key condition.

Q66. What is the primary purpose of mutual information?

A. Mean only. B. Variance only. C. Pearson correlation only. D. Measures statistical dependence between feature and target.

Select an answer to check.

Answer: Measures statistical dependence between feature and target.

Here, Measures statistical dependence between feature and target. is the right choice. Captures non-linear relationships. It fits the requirement in the prompt about what is the primary purpose of mutual information. Competing choices sound plausible, but they miss the key condition.

Q67. Which statement about mutual information is most accurate?

A. Measures statistical dependence between feature and target. B. Mean only. C. Variance only. D. Pearson correlation only.

Select an answer to check.

Answer: Measures statistical dependence between feature and target.

In this case, Measures statistical dependence between feature and target. is correct. Captures non-linear relationships. It fits the requirement in the prompt about which statement about mutual information is most accurate. Competing choices sound plausible, but they miss the key condition.

Q68. How is mutual information best characterized?

A. Variance only. B. Mean only. C. Measures statistical dependence between feature and target. D. Pearson correlation only.

Select an answer to check.

Answer: Measures statistical dependence between feature and target.

The best option here is Measures statistical dependence between feature and target.. Captures non-linear relationships. It fits the requirement in the prompt about how is mutual information best characterized. Competing choices sound plausible, but they miss the key condition.

Q69. Which option best describes variance threshold?

A. Drop rare categories only. B. Drop near-constant features. C. Drop categorical features. D. Drop high-variance features.

Select an answer to check.

Answer: Drop near-constant features.

For this question, Drop near-constant features. is correct. Cheap simple filter. It fits the requirement in the prompt about which option best describes variance threshold. Competing choices sound plausible, but they miss the key condition.

Q70. What is the primary purpose of variance threshold?

A. Drop near-constant features. B. Drop categorical features. C. Drop rare categories only. D. Drop high-variance features.

Select an answer to check.

Answer: Drop near-constant features.

Drop near-constant features. is the correct answer here. Cheap simple filter. It fits the requirement in the prompt about what is the primary purpose of variance threshold. Competing choices sound plausible, but they miss the key condition.

Q71. Which statement about variance threshold is most accurate?

A. Drop near-constant features. B. Drop high-variance features. C. Drop categorical features. D. Drop rare categories only.

Select an answer to check.

Answer: Drop near-constant features.

Here, Drop near-constant features. is the right choice. Cheap simple filter. This is the most accurate statement for which statement about variance threshold is most accurate. Competing choices sound plausible, but they miss the key condition.

Q72. How is variance threshold best characterized?

A. Drop high-variance features. B. Drop rare categories only. C. Drop categorical features. D. Drop near-constant features.

Select an answer to check.

Answer: Drop near-constant features.

In this case, Drop near-constant features. is correct. Cheap simple filter. This is the most accurate statement for how is variance threshold best characterized. Competing choices sound plausible, but they miss the key condition.

Q73. Which option best describes multicollinearity?

A. High correlation among features. B. Class imbalance. C. Concept drift. D. Data leakage.

Select an answer to check.

Answer: High correlation among features.

The best option here is High correlation among features.. Hurts linear model interpretability. This is the most accurate statement for which option best describes multicollinearity. Competing choices sound plausible, but they miss the key condition.

Q74. What is the primary purpose of multicollinearity?

A. Data leakage. B. Concept drift. C. High correlation among features. D. Class imbalance.

Select an answer to check.

Answer: High correlation among features.

For this question, High correlation among features. is correct. Hurts linear model interpretability. This is the most accurate statement for what is the primary purpose of multicollinearity. Competing choices sound plausible, but they miss the key condition.

Q75. Which statement about multicollinearity is most accurate?

A. Data leakage. B. Class imbalance. C. High correlation among features. D. Concept drift.

Select an answer to check.

Answer: High correlation among features.

High correlation among features. is the correct answer here. Hurts linear model interpretability. This is the most accurate statement for which statement about multicollinearity is most accurate. Competing choices sound plausible, but they miss the key condition.

Q76. How is multicollinearity best characterized?

A. Class imbalance. B. Data leakage. C. High correlation among features. D. Concept drift.

Select an answer to check.

Answer: High correlation among features.

Here, High correlation among features. is the right choice. Hurts linear model interpretability. It aligns directly with what the question asks about how is multicollinearity best characterized. The remaining choices fail because they don’t satisfy the full definition.

Q77. Which option best describes feature interactions?

A. Pure aggregation only. B. Pure embeddings only. C. Combinations of features that matter together. D. Single-feature transforms only.

Select an answer to check.

Answer: Combinations of features that matter together.

In this case, Combinations of features that matter together. is correct. Trees/NNs find these implicitly. It aligns directly with what the question asks about which option best describes feature interactions. The remaining choices fail because they don’t satisfy the full definition.

Q78. What is the primary purpose of feature interactions?

A. Pure aggregation only. B. Single-feature transforms only. C. Pure embeddings only. D. Combinations of features that matter together.

Select an answer to check.

Answer: Combinations of features that matter together.

The best option here is Combinations of features that matter together.. Trees/NNs find these implicitly. It aligns directly with what the question asks about what is the primary purpose of feature interactions. The remaining choices fail because they don’t satisfy the full definition.

Q79. Which statement about feature interactions is most accurate?

A. Pure aggregation only. B. Pure embeddings only. C. Combinations of features that matter together. D. Single-feature transforms only.

Select an answer to check.

Answer: Combinations of features that matter together.

For this question, Combinations of features that matter together. is correct. Trees/NNs find these implicitly. It aligns directly with what the question asks about which statement about feature interactions is most accurate. The remaining choices fail because they don’t satisfy the full definition.

Q80. How is feature interactions best characterized?

A. Pure aggregation only. B. Pure embeddings only. C. Combinations of features that matter together. D. Single-feature transforms only.

Select an answer to check.

Answer: Combinations of features that matter together.

Combinations of features that matter together. is the correct answer here. Trees/NNs find these implicitly. It aligns directly with what the question asks about how is feature interactions best characterized. The remaining choices fail because they don’t satisfy the full definition.

Q81. Which option best describes data leakage?

A. Information from the future/test bleeding into training. B. Class imbalance. C. High variance features. D. Outliers in features.

Select an answer to check.

Answer: Information from the future/test bleeding into training.

Here, Information from the future/test bleeding into training. is the right choice. Causes inflated metrics. This matches the core idea being tested around which option best describes data leakage. The remaining choices fail because they don’t satisfy the full definition.

Q82. What is the primary purpose of data leakage?

A. Class imbalance. B. High variance features. C. Outliers in features. D. Information from the future/test bleeding into training.

Select an answer to check.

Answer: Information from the future/test bleeding into training.

In this case, Information from the future/test bleeding into training. is correct. Causes inflated metrics. This matches the core idea being tested around what is the primary purpose of data leakage. The remaining choices fail because they don’t satisfy the full definition.

Q83. Which statement about data leakage is most accurate?

A. Class imbalance. B. Outliers in features. C. High variance features. D. Information from the future/test bleeding into training.

Select an answer to check.

Answer: Information from the future/test bleeding into training.

The best option here is Information from the future/test bleeding into training.. Causes inflated metrics. This matches the core idea being tested around which statement about data leakage is most accurate. The remaining choices fail because they don’t satisfy the full definition.

Q84. How is data leakage best characterized?

A. High variance features. B. Class imbalance. C. Outliers in features. D. Information from the future/test bleeding into training.

Select an answer to check.

Answer: Information from the future/test bleeding into training.

For this question, Information from the future/test bleeding into training. is correct. Causes inflated metrics. This matches the core idea being tested around how is data leakage best characterized. The remaining choices fail because they don’t satisfy the full definition.

Q85. Which option best describes temporal split?

A. Use time-based train/val/test splits. B. Stratified split ignoring time. C. K-fold ignoring time. D. Random split for time series.

Select an answer to check.

Answer: Use time-based train/val/test splits.

Use time-based train/val/test splits. is the correct answer here. Prevents temporal leakage. This matches the core idea being tested around which option best describes temporal split. The remaining choices fail because they don’t satisfy the full definition.

Q86. What is the primary purpose of temporal split?

A. K-fold ignoring time. B. Random split for time series. C. Stratified split ignoring time. D. Use time-based train/val/test splits.

Select an answer to check.

Answer: Use time-based train/val/test splits.

Here, Use time-based train/val/test splits. is the right choice. Prevents temporal leakage. That is exactly the concept behind what is the primary purpose of temporal split in this context. The remaining choices fail because they don’t satisfy the full definition.

Q87. Which statement about temporal split is most accurate?

A. Stratified split ignoring time. B. Random split for time series. C. Use time-based train/val/test splits. D. K-fold ignoring time.

Select an answer to check.

Answer: Use time-based train/val/test splits.

In this case, Use time-based train/val/test splits. is correct. Prevents temporal leakage. That is exactly the concept behind which statement about temporal split is most accurate in this context. The remaining choices fail because they don’t satisfy the full definition.

Q88. How is temporal split best characterized?

A. Random split for time series. B. Use time-based train/val/test splits. C. K-fold ignoring time. D. Stratified split ignoring time.

Select an answer to check.

Answer: Use time-based train/val/test splits.

The best option here is Use time-based train/val/test splits.. Prevents temporal leakage. That is exactly the concept behind how is temporal split best characterized in this context. The remaining choices fail because they don’t satisfy the full definition.

Q89. Which option best describes feature store?

A. A logging service. B. A vector index only. C. Centralized service for storing/serving features. D. A retraining queue.

Select an answer to check.

Answer: Centralized service for storing/serving features.

For this question, Centralized service for storing/serving features. is correct. Reduces train/serve skew. That is exactly the concept behind which option best describes feature store in this context. The remaining choices fail because they don’t satisfy the full definition.

Q90. What is the primary purpose of feature store?

A. A logging service. B. A vector index only. C. A retraining queue. D. Centralized service for storing/serving features.

Select an answer to check.

Answer: Centralized service for storing/serving features.

Centralized service for storing/serving features. is the correct answer here. Reduces train/serve skew. That is exactly the concept behind what is the primary purpose of feature store in this context. The remaining choices fail because they don’t satisfy the full definition.

Q91. Which statement about feature store is most accurate?

A. A logging service. B. A retraining queue. C. A vector index only. D. Centralized service for storing/serving features.

Select an answer to check.

Answer: Centralized service for storing/serving features.

Here, Centralized service for storing/serving features. is the right choice. Reduces train/serve skew. It fits the requirement in the prompt about which statement about feature store is most accurate. The remaining choices fail because they don’t satisfy the full definition.

Q92. How is feature store best characterized?

A. Centralized service for storing/serving features. B. A vector index only. C. A logging service. D. A retraining queue.

Select an answer to check.

Answer: Centralized service for storing/serving features.

In this case, Centralized service for storing/serving features. is correct. Reduces train/serve skew. It fits the requirement in the prompt about how is feature store best characterized. The remaining choices fail because they don’t satisfy the full definition.

Q93. Which option best describes train/serve skew?

A. Outlier presence. B. Concept drift. C. Difference between training-time and serving-time features. D. Class imbalance.

Select an answer to check.

Answer: Difference between training-time and serving-time features.

The best option here is Difference between training-time and serving-time features.. Causes silent quality regressions. It fits the requirement in the prompt about which option best describes train/serve skew. The remaining choices fail because they don’t satisfy the full definition.

Q94. What is the primary purpose of train/serve skew?

A. Outlier presence. B. Concept drift. C. Difference between training-time and serving-time features. D. Class imbalance.

Select an answer to check.

Answer: Difference between training-time and serving-time features.

For this question, Difference between training-time and serving-time features. is correct. Causes silent quality regressions. It fits the requirement in the prompt about what is the primary purpose of train/serve skew. The remaining choices fail because they don’t satisfy the full definition.

Q95. Which statement about train/serve skew is most accurate?

A. Outlier presence. B. Concept drift. C. Class imbalance. D. Difference between training-time and serving-time features.

Select an answer to check.

Answer: Difference between training-time and serving-time features.

Difference between training-time and serving-time features. is the correct answer here. Causes silent quality regressions. It fits the requirement in the prompt about which statement about train/serve skew is most accurate. The remaining choices fail because they don’t satisfy the full definition.

Q96. How is train/serve skew best characterized?

A. Concept drift. B. Difference between training-time and serving-time features. C. Outlier presence. D. Class imbalance.

Select an answer to check.

Answer: Difference between training-time and serving-time features.

Here, Difference between training-time and serving-time features. is the right choice. Causes silent quality regressions. This is the most accurate statement for how is train/serve skew best characterized. The remaining choices fail because they don’t satisfy the full definition.

Q97. Which option best describes text features (TF-IDF)?

A. Token frequency reweighted by inverse doc frequency. B. Pure embeddings only. C. Hashing only. D. Pure word counts only.

Select an answer to check.

Answer: Token frequency reweighted by inverse doc frequency.

In this case, Token frequency reweighted by inverse doc frequency. is correct. Strong baseline for classical text. This is the most accurate statement for which option best describes text features (tf-idf). The remaining choices fail because they don’t satisfy the full definition.

Q98. What is the primary purpose of text features (TF-IDF)?

A. Pure word counts only. B. Pure embeddings only. C. Token frequency reweighted by inverse doc frequency. D. Hashing only.

Select an answer to check.

Answer: Token frequency reweighted by inverse doc frequency.

The best option here is Token frequency reweighted by inverse doc frequency.. Strong baseline for classical text. This is the most accurate statement for what is the primary purpose of text features. The remaining choices fail because they don’t satisfy the full definition.

Q99. Which statement about text features (TF-IDF) is most accurate?

A. Hashing only. B. Pure embeddings only. C. Pure word counts only. D. Token frequency reweighted by inverse doc frequency.

Select an answer to check.

Answer: Token frequency reweighted by inverse doc frequency.

For this question, Token frequency reweighted by inverse doc frequency. is correct. Strong baseline for classical text. This is the most accurate statement for which statement about text features (tf-idf) is most. The remaining choices fail because they don’t satisfy the full definition.

Q100. How is text features (TF-IDF) best characterized?

A. Token frequency reweighted by inverse doc frequency. B. Hashing only. C. Pure word counts only. D. Pure embeddings only.

Select an answer to check.

Answer: Token frequency reweighted by inverse doc frequency.

Token frequency reweighted by inverse doc frequency. is the correct answer here. Strong baseline for classical text. This is the most accurate statement for how is text features (tf-idf) best characterized. The remaining choices fail because they don’t satisfy the full definition.