MCQs on Data & Feature Engineering

1. Data vs. Information

1. What is the key difference between data and information?

o a) Data is processed, whereas information is raw

o b) Data is raw facts, whereas information is processed data ✅

o c) Data is always numerical, whereas information is textual

o d) There is no difference

2. Which of the following best describes information?

o a) Unorganized raw numbers

o b) Processed data that is meaningful ✅

o c) A collection of random symbols

o d) All of the above

3. In AI, what is an essential step before converting data into information?

o a) Preprocessing and cleaning ✅

o b) Ignoring missing values

o c) Using only numerical data

o d) None of the above

2. Types of Data

4. Which of the following is an example of numerical data?

o a) Customer reviews

o b) Temperature readings ✅

o c) Social media posts

o d) Gender

5. Discrete numerical data represents:

o a) Continuous values
 o b) Countable values ✅

o c) Infinite possible values

o d) All of the above

6. What type of data is "Age in years"?

o a) Discrete numerical ✅

o b) Continuous numerical

o c) Ordinal

o d) Nominal

7. Which of the following is an example of continuous numerical data?

o a) Number of students in a class

o b) Height of a person ✅

o c) Zip code

o d) Gender

8. What type of data is "Education level (Primary, Secondary, Tertiary)"?

o a) Nominal

o b) Ordinal ✅

o c) Discrete

o d) Continuous

9. Which of the following is an example of categorical data?

o a) Blood pressure readings

o b) Eye color ✅

o c) Number of books sold

o d) Temperature in Celsius

10. Which category does time series data fall into?

 a) Discrete numerical

 b) Continuous numerical ✅
  c) Nominal

 d) Unstructured

3. Data Labeling

11. What is data labeling?

 a) Process of assigning categories to unstructured data

 b) Process of manually tagging data for supervised learning ✅

 c) A method to remove outliers

 d) None of the above

12. Labeled data is essential for:

 a) Supervised learning ✅

 b) Unsupervised learning

 c) Reinforcement learning

 d) Clustering

13. Which of the following is NOT a challenge in data labeling?

 a) Cost and time

 b) Labeling errors

 c) Data duplication ✅

 d) Subjectivity in labeling

4. Feature and Feature Selection

14. In machine learning, what is a feature?

 a) An attribute or characteristic of data ✅

 b) A type of model

 c) A type of supervised learning algorithm

 d) None of the above

 a) Reduces overfitting ✅

 b) Improves model accuracy ✅

 c) Enhances model interpretability ✅

 d) All of the above ✅

16. Which of the following is NOT a feature selection method?

 a) Sequential forward selection

 b) Data augmentation ✅

 c) Bidirectional feature selection

 d) Sequential backward selection

5. Feature Selection Algorithms

17. What is the goal of feature selection?

 a) To increase model complexity

 b) To select the most relevant features while reducing redundancy ✅

 c) To add more irrelevant features

 d) To increase the number of features

18. Which algorithm is used in feature selection?

 a) Recursive Feature Elimination (RFE) ✅

 b) K-Means

 c) Decision Trees

 d) Random Forest

19. Which feature selection method starts with an empty feature set and adds features one
by one?

 a) Sequential Forward Selection (SFS) ✅

 b) Sequential Backward Selection

 d) Genetic Algorithm

20. In Sequential Backward Selection (SBS), how are features removed?

 a) One by one starting with the most important feature

 b) One by one starting with the least important feature ✅

 c) Randomly

 d) Based on feature correlation

21. What is bidirectional feature selection?

 a) Selecting features randomly

 b) A combination of forward and backward selection methods ✅

 c) A supervised learning algorithm

 d) None of the above

6. Feature Extraction

22. What is feature extraction?

 a) Transforming raw data into a set of meaningful features ✅

 b) Randomly selecting features

 c) Removing missing values

 d) None of the above

23. Principal Component Analysis (PCA) is used for:

 a) Feature extraction ✅

 b) Feature selection

 c) Data labeling

 d) Data augmentation

24. What is the main advantage of feature extraction?

 a) Reduces dimensionality while preserving important information ✅

 c) Adds irrelevant features

 d) None of the above

25. Which of the following is NOT a feature extraction technique?

 a) Principal Component Analysis

 b) Linear Discriminant Analysis

 c) K-Means Clustering ✅

 d) Autoencoders

More MCQs

26. Which type of data is commonly used in time-series analysis?

 a) Continuous numerical ✅

 b) Discrete numerical

 c) Ordinal

 d) Nominal

27. Which machine learning approach benefits the most from labeled data?

 a) Supervised Learning ✅

 b) Unsupervised Learning

 c) Reinforcement Learning

 d) Semi-supervised Learning

28. What is a major challenge in feature selection?

 a) Computational cost for high-dimensional data ✅

 b) Low memory usage

 c) Lack of machine learning models

 d) Removing missing values

29. What is the role of correlation in feature selection?

 b) Increases computational complexity

 c) Reduces model accuracy

 d) None of the above

30. Feature engineering is important because:

 a) It improves model performance and interpretability ✅

 b) It reduces the need for training

 c) It increases data redundancy

 d) It is not necessary in AI

31. Which of the following is an example of ordinal data?

 a) Customer satisfaction ratings (e.g., Poor, Average, Good, Excellent) ✅

 b) Eye color

 c) Credit card number

 d) Phone number

32. In data preprocessing, what is the purpose of normalization?

 a) To scale data within a specific range (e.g., 0 to 1 or -1 to 1) ✅

 b) To increase the dimensionality of data

 c) To remove missing values

 d) To replace categorical variables

33. One-hot encoding is commonly used for:

 a) Converting categorical data into numerical format ✅

 b) Normalization

 c) Data cleaning

 d) Handling missing values

34. Which method is commonly used to handle missing data?

 a) Imputation with mean, median, or mode ✅

 c) Ignoring missing values

 d) Feature scaling

32. Data Labeling & Supervised Learning

35. Data labeling is crucial for which type of machine learning?

 a) Supervised learning ✅

 b) Unsupervised learning

 c) Reinforcement learning

 d) Self-learning

36. What is weakly labeled data?

 a) Data that has labels generated automatically with some errors ✅

 b) Data with no labels

 c) Data with perfect manual labeling

 d) Completely irrelevant data

37. Which AI field benefits the most from semi-supervised learning?

 a) Medical image classification (few labeled images available) ✅

 b) Clustering algorithms

 c) Reinforcement learning in games

 d) Search engine indexing

33. Feature Engineering & Selection

38. What is the main purpose of feature selection?

 a) To remove redundant or irrelevant features to improve model performance ✅

 b) To add more features to improve accuracy

 c) To reduce training time by increasing dimensionality

39. Recursive Feature Elimination (RFE) works by:

 a) Removing the least important feature iteratively ✅

 b) Adding new features iteratively

 c) Clustering the most important features

 d) Selecting features randomly

40. What is feature extraction mainly used for?

 a) Transforming raw data into meaningful features ✅

 b) Increasing model complexity

 c) Ignoring missing data

 d) Converting numerical data to categorical

41. Which of the following is an advantage of feature extraction?

 a) Reduces dimensionality and computational cost ✅

 b) Increases the number of features

 c) Increases overfitting

 d) Makes the dataset more complex

34. Feature Selection Techniques

42. What is a filter-based feature selection method?

 a) A technique that selects features based on statistical tests ✅

 b) A technique that selects features based on model performance

 c) A method that selects features randomly

 d) A method that removes missing values

43. Information Gain is commonly used in:

 a) Feature selection for decision trees ✅

 b) Feature extraction
  c) Clustering algorithms

 d) Neural networks

44. Which feature selection method ranks features based on correlation?

 a) Pearson’s correlation coefficient ✅

 b) Principal Component Analysis

 c) K-Means Clustering

 d) Autoencoders

45. The curse of dimensionality refers to:

 a) A situation where too many features reduce model performance ✅

 b) When data preprocessing fails

 c) When supervised learning is not applicable

 d) A type of machine learning model

35. Feature Engineering in AI

46. Which technique is used for feature engineering in deep learning?

 a) Autoencoders ✅

 b) Decision trees

 c) One-hot encoding

 d) Pearson correlation

47. Why is dimensionality reduction important in machine learning?

 a) To remove noise and improve computational efficiency ✅

 b) To increase feature space

 c) To slow down training speed

 d) To reduce model accuracy

48. In text-based AI models, what is the most commonly used feature extraction technique?

 a) TF-IDF (Term Frequency-Inverse Document Frequency) ✅

 c) Data augmentation

 d) Variance thresholding

49. Principal Component Analysis (PCA) works by:

 a) Finding new axes that maximize variance while reducing dimensions ✅

 b) Removing missing values

 c) Sorting features alphabetically

 d) Selecting the most correlated features

50. Which machine learning models automatically perform feature selection?

 a) Decision Trees & Random Forest ✅

 b) K-Means Clustering

 c) Neural Networks

 d) Principal Component Analysis

51. Which of the following is an example of unstructured data?

 a) Social media posts ✅

 b) Temperature readings

 c) Customer age data

 d) ZIP codes

52. Which technique is used to handle outliers in a dataset?

 a) Winsorization (Capping extreme values) ✅

 b) Min-Max Scaling

 c) One-hot encoding

 d) Removing missing values

53. Which of the following statements is true about categorical data?

 a) Categorical data can be nominal or ordinal ✅

 b) Categorical data must always be numerical

 d) Ordinal data has no meaningful order

54. What is the main problem when working with imbalanced datasets?

 a) The model may be biased toward the majority class ✅

 b) The dataset contains too many features

 c) Missing values increase

 d) Feature selection becomes impossible

37. Data Labeling & Feature Engineering

55. In supervised learning, the quality of data labels directly affects:

 a) Model performance and accuracy ✅

 b) Feature extraction only

 c) The number of missing values

 d) Unsupervised learning models

56. A dataset with multiple dependent variables is known as:

 a) Multivariate data ✅

 b) Bivariate data

 c) Univariate data

 d) Structured data

57. Which of the following is a disadvantage of manual data labeling?

 a) Time-consuming and expensive ✅

 b) Not applicable to supervised learning

 c) Requires high computational power

 d) Unnecessary for high-dimensional data

58. Which term describes the process of generating new labeled training data from existing
data?
 a) Data augmentation ✅

 b) Feature selection

 c) Principal Component Analysis

 d) Clustering

38. Feature Selection & Feature Extraction

59. The primary goal of feature selection is to:

 a) Identify the most relevant features while removing redundant ones ✅

 b) Increase the number of features in a dataset

 c) Reduce the training set size

 d) Convert numerical data into categorical data

60. What is the difference between feature selection and feature extraction?

 a) Feature selection removes redundant features, whereas feature extraction

 b) Feature selection is only for categorical data

 c) Feature extraction is used in classification but not regression

 d) Both are the same

61. Which algorithm can perform automatic feature selection?

 a) LASSO Regression ✅

 b) K-Means Clustering

 c) K-Nearest Neighbors

 d) Principal Component Analysis

62. Which of the following is NOT a feature selection method?

 a) Cross-validation ✅

 b) Recursive Feature Elimination

 c) Information Gain
 d) Mutual Information

39. Feature Engineering in Machine Learning

63. What is one of the key benefits of feature engineering?

 a) Improves model performance by creating meaningful features ✅

 b) Increases overfitting risk

 c) Reduces dataset size

 d) Prevents data cleaning

64. When should feature scaling be applied?

 a) When using distance-based algorithms like KNN and SVM ✅

 b) When working with categorical data

 c) Only when dealing with large datasets

 d) Feature scaling is not necessary in machine learning

65. What is an important step when performing text feature engineering?

 a) Tokenization and vectorization ✅

 b) One-hot encoding

 c) Feature scaling

 d) Min-Max Normalization

40. Advanced Feature Engineering Techniques

66. Principal Component Analysis (PCA) is most useful when:

 a) There are many correlated features in a dataset ✅

 b) The dataset is already well-structured

 c) We want to perform data labeling

 d) The dataset has missing values

67. What is the main purpose of t-SNE (t-Distributed Stochastic Neighbor Embedding)?
 a) Dimensionality reduction for data visualization ✅

 b) Feature selection

 c) Text classification

 d) Clustering

68. Which feature selection method is commonly used for handling high-dimensional data?

 a) LASSO Regression ✅

 b) K-Means

 c) Decision Trees

 d) SVM

69. In time series feature engineering, which of the following is a commonly used feature?

 a) Rolling mean ✅

 b) One-hot encoding

 c) Word embeddings

 d) Data augmentation

41. Feature Engineering in Deep Learning & AI

70. How does deep learning handle feature engineering?

 a) It learns features automatically from raw data ✅

 b) Requires manual feature selection

 c) Uses predefined statistical tests

 d) Ignores feature extraction