Gyansetu Team 
Sharpen Your Concepts for Job Interviews, Assessments & Real-World ML Applications
Machine Learning (ML) is no longer just a buzzword — it’s the backbone of smart technologies across industries. Whether you’re applying for roles like ML Engineer, Data Scientist, AI Researcher, or even Data Analyst, you must be equipped with a strong understanding of ML fundamentals and applied concepts.
Section 1: Basics of Machine Learning
Q1. What is Machine Learning?
a) A way to clean data
b) A form of database management
c) A subset of AI that enables systems to learn from data
d) Writing code manually
Answer: c) A subset of AI that enables systems to learn from data
Q2. Which of the following is not a type of machine learning?
a) Supervised
b) Unsupervised
c) Reinforcement
d) Constructive
Answer: d) Constructive
Q3. What is the first step in a machine learning workflow?
a) Model selection
b) Data collection
c) Hyperparameter tuning
d) Visualization
Answer: b) Data collection
Q4. Which type of data does machine learning rely on?
a) Static data only
b) Historical and real-time data
c) Only structured data
d) Only images and videos
Answer: b) Historical and real-time data
Q5. In machine learning, what is ‘training data’?
a) Data used to fine-tune parameters
b) Data used to test accuracy
c) Data used to build the model
d) Data stored in memory
Answer: c) Data used to build the model
Q6. Which of the following is the main goal of machine learning?
a) Extracting data
b) Automating decision making
c) Encrypting data
d) Storing large volumes of data
Answer: b) Automating decision making
Q7. Which term refers to an ML system improving over time without being explicitly programmed?
a) Static coding
b) Hardcoded loops
c) Learning
d) Debugging
Answer: c) Learning
Learn to Predict the Future — Master Machine Learning Today!
Section 2: Supervised Learning
Q8. Which of the following is a supervised learning task?
a) Clustering
b) Classification
c) Association Rule Mining
d) Dimensionality Reduction
Answer: b) Classification
Q9. What is the target in supervised learning?
a) Not needed
b) A label or output variable
c) A cluster center
d) A distance function
Answer: b) A label or output variable
Q10. Regression tasks are used to:
a) Predict classes
b) Reduce dimensionality
c) Predict continuous values
d) Count categories
Answer: c) Predict continuous values
Q11. Which algorithm is used for classification tasks?
a) K-Means
b) Linear Regression
c) Decision Tree
d) PCA
Answer: c) Decision Tree
Q12. Which one is a linear model?
a) SVM with RBF kernel
b) Naive Bayes
c) Logistic Regression
d) Random Forest
Answer: c) Logistic Regression
Q13. In supervised learning, accuracy is used to:
a) Increase runtime
b) Evaluate model performance
c) Sort the data
d) Encrypt the dataset
Answer: b) Evaluate model performance
Q14. What is the main disadvantage of supervised learning?
a) Needs a large amount of unlabeled data
b) Cannot handle real-time data
c) Requires labeled data
d) Needs GPU always
Answer: c) Requires labeled data
Section 3: Unsupervised Learning
Q15. What is the primary goal of unsupervised learning?
a) Predict outcomes
b) Find patterns and groupings
c) Optimize weights
d) Generate code
Answer: b) Find patterns and groupings
Q16. Which of these is an unsupervised learning algorithm?
a) K-Means
b) Decision Trees
c) SVM
d) Logistic Regression
Answer: a) K-Means
Q17. What is clustering?
a) Predicting labels
b) Dividing data into categories
c) Grouping data points based on similarity
d) Sorting variables
Answer: c) Grouping data points based on similarity
Q18. Which of the following is not a clustering algorithm?
a) K-Means
b) DBSCAN
c) PCA
d) Hierarchical Clustering
Answer: c) PCA
Q19. Dimensionality reduction is used for:
a) Clustering
b) Visualization and performance improvement
c) Hyperparameter tuning
d) Regression
Answer: b) Visualization and performance improvement
Q20. Principal Component Analysis (PCA) helps with:
a) Classification
b) Feature selection
c) Dimensionality reduction
d) Regression
Answer: c) Dimensionality reduction
Q21. Association rule mining is used to:
a) Train models
b) Discover relationships between variables
c) Predict categories
d) Normalize datasets
Answer: b) Discover relationships between variables
Section 4: Model Evaluation & Metrics
Q22. What does a confusion matrix measure?
a) Model runtime
b) Misclassification
c) Execution cost
d) Correlation
Answer: b) Misclassification
Q23. Precision is defined as:
a) TP / (TP + FP)
b) TP / (TP + FN)
c) (TP + TN) / Total
d) FP / Total
Answer: a) TP / (TP + FP)
Q24. What is the ROC curve used for?
a) Measuring variance
b) Evaluating binary classification models
c) Finding clusters
d) Scaling values
Answer: b) Evaluating binary classification models
Q25. Which metric is ideal for imbalanced datasets?
a) Accuracy
b) Recall
c) F1 Score
d) MAE
Answer: c) F1 Score
From Data to Decisions: Your ML Journey Starts Here.
Q26. MAE stands for:
a) Mean Absolute Error
b) Mean Accuracy Estimation
c) Maximum Accuracy Estimate
d) Model Average Estimation
Answer: a) Mean Absolute Error
Q27. R-squared value is used in:
a) Classification
b) Regression
c) Clustering
d) NLP
Answer: b) Regression
Q28. Cross-validation helps to:
a) Increase dataset
b) Estimate model generalization
c) Store output
d) Encrypt labels
Answer: b) Estimate model generalization
Section 5: Overfitting & Underfitting
Q29. Overfitting occurs when:
a) Model performs well on new data
b) Model is simple
c) Model memorizes training data
d) Dataset is large
Answer: c) Model memorizes training data
Q30. Underfitting leads to:
a) Poor training and test accuracy
b) High test accuracy
c) High variance
d) Large confusion matrix
Answer: a) Poor training and test accuracy
Q31. Which technique helps prevent overfitting?
a) Increasing learning rate
b) Cross-validation
c) Using small datasets
d) Ignoring outliers
Answer: b) Cross-validation
Q32. Which of these is a regularization technique?
a) Feature scaling
b) L1 and L2
c) Dropout
d) Binning
Answer: b) L1 and L2
Q33. Dropout is used in:
a) Clustering
b) Deep Learning
c) Regression
d) Association mining
Answer: b) Deep Learning
Q34. High variance means:
a) Underfitting
b) Data leakage
c) Overfitting
d) Noise
Answer: c) Overfitting
Q35. Bias is related to:
a) Overfitting
b) Underfitting
c) Hyperparameter tuning
d) Gradient clipping
Answer: b) Underfitting
Section 6: Algorithms & Models
Q36. Which algorithm is best for predicting a continuous numerical value?
a) K-Means
b) Linear Regression
c) Naive Bayes
d) Decision Tree (Classifier)
Answer: b) Linear Regression
Q37. Which algorithm is known for using “if-else” logic?
a) Logistic Regression
b) Decision Tree
c) K-Means
d) SVM
Answer: b) Decision Tree
Q38. SVM stands for:
a) Sequential Vector Machine
b) Sample Variance Model
c) Support Vector Machine
d) Supervised Variable Model
Answer: c) Support Vector Machine
Q39. Random Forest is an ensemble of:
a) Neural Networks
b) Logistic Models
c) Decision Trees
d) Linear Equations
Answer: c) Decision Trees
Q40. KNN stands for:
a) Kernel Neural Network
b) K-Nearest Neighbors
c) Key Numeric Network
d) Known Node Network
Answer: b) K-Nearest Neighbors
Q41. What is a hyperparameter in ML?
a) A type of label
b) A fixed setting like learning rate or tree depth
c) Output of the model
d) An evaluation metric
Answer: b) A fixed setting like learning rate or tree depth
Q42. Which ML model is commonly used in spam email classification?
a) Linear Regression
b) K-Means
c) Naive Bayes
d) PCA
Answer: c) Naive Bayes
Section 7: Feature Engineering
Q43. Feature engineering is the process of:
a) Scaling predictions
b) Creating new input features from raw data
c) Optimizing model weights
d) Visualizing loss functions
Answer: b) Creating new input features from raw data
Q44. One-Hot Encoding is used for:
a) Normalizing numerical values
b) Splitting data
c) Converting categorical variables
d) Creating histograms
Answer: c) Converting categorical variables
Q45. What is feature scaling?
a) Dropping rows with missing values
b) Reducing model complexity
c) Rescaling features to a standard range
d) Balancing class labels
Answer: c) Rescaling features to a standard range
Q46. Which method is used to handle missing values?
a) PCA
b) KNN
c) Imputation
d) Normalization
Answer: c) Imputation
Q47. Feature selection helps with:
a) Removing irrelevant features
b) Increasing data noise
c) Decreasing accuracy
d) Encoding labels
Answer: a) Removing irrelevant features
Q48. A correlation matrix is used to:
a) Evaluate model performance
b) Tune hyperparameters
c) Identify relationships between features
d) Predict outcomes
Answer: c) Identify relationships between features
Q49. Polynomial features are created to:
a) Normalize data
b) Increase complexity for better fit
c) Reduce dimensionality
d) Visualize clusters
Answer: b) Increase complexity for better fit
Section 8: Python & ML Libraries
Q50. Which Python library is used primarily for machine learning?
a) NumPy
b) Pandas
c) scikit-learn
d) Flask
Answer: c) scikit-learn
Smart Algorithms, Smarter Careers
Q51. Pandas is mainly used for:
a) Web development
b) Image processing
c) Data manipulation and analysis
d) Deep learning
Answer: c) Data manipulation and analysis
Q52. NumPy is essential for:
a) Plotting graphs
b) Statistical models
c) Numerical computations
d) Label encoding
Answer: c) Numerical computations
Q53. Which library is widely used for deep learning in Python?
a) Matplotlib
b) TensorFlow
c) OpenCV
d) BeautifulSoup
Answer: b) TensorFlow
Q54. What does .fit() do in scikit-learn models?
a) Tests the model
b) Stores the result
c) Trains the model
d) Predicts new data
Answer: c) Trains the model
Q55. Which command is used to install scikit-learn?
a) pip install sklearn
b) install scikit
c) pip install pandas
d) sklearn install
Answer: a) pip install sklearn
Q56. Matplotlib is used for:
a) Data modeling
b) Text preprocessing
c) Data visualization
d) Web scraping
Answer: c) Data visualization
Section 9: Deep Learning
Q57. What is deep learning a subset of?
a) Web programming
b) Data structures
c) Machine Learning
d) Operating systems
Answer: c) Machine Learning
Q58. Which of the following is a deep learning framework?
a) Django
b) TensorFlow
c) Flask
d) Pytest
Answer: b) TensorFlow
Q59. What is a perceptron?
a) Data visualization tool
b) Basic unit of a neural network
c) Decision tree variant
d) File format
Answer: b) Basic unit of a neural network
Q60. CNNs are mainly used for:
a) Time-series prediction
b) Image data
c) Database management
d) Speech-to-text
Answer: b) Image data
Q61. RNNs are best suited for:
a) Tabular data
b) Text and sequences
c) Clustering
d) Regression
Answer: b) Text and sequences
Q62. What does backpropagation do?
a) Randomly updates weights
b) Propagates errors to adjust weights
c) Splits the dataset
d) Encrypts model
Answer: b) Propagates errors to adjust weights
Q63. The vanishing gradient problem occurs in:
a) Decision Trees
b) Random Forest
c) Deep Neural Networks
d) Linear Regression
Answer: c) Deep Neural Networks
Section 10: Miscellaneous & Real-World Use Cases
Q64. Which company popularized the use of ML in product recommendations?
a) Walmart
b) Amazon
c) Toyota
d) IBM
Answer: b) Amazon
Q65. Credit card fraud detection uses which ML approach?
a) Clustering
b) Classification
c) Dimensionality reduction
d) Regression
Answer: b) Classification
Q66. What’s a common use case for clustering in real life?
a) Weather prediction
b) Customer segmentation
c) Loan approval
d) Image classification
Answer: b) Customer segmentation
Q67. Which ML task is best for predicting house prices?
a) Classification
b) Clustering
c) Regression
d) Reinforcement Learning
Answer: c) Regression
Q68. Which term describes a model that performs well on training data but poorly on test data?
a) Underfit
b) Overfit
c) Balanced
d) Tuned
Answer: b) Overfit
Q69. Reinforcement learning is based on:
a) Historical data
b) Trial and error
c) Image classification
d) Real-time streaming
Answer: b) Trial and error
Q70. Chatbots commonly use which combination of techniques?
a) Linear regression and DBSCAN
b) Clustering and PCA
c) NLP and Machine Learning
d) Decision Trees and SQL
Answer: c) NLP and Machine Learning
Conclusion
Machine Learning is a vast and evolving field. These 70 MCQs cover the core fundamentals, popular algorithms, evaluation metrics, deep learning basics, and real-world use cases to help you prepare for:
- Job interviews
- Online assessments or quizzes
- Skill refreshers and certifications
Don’t just memorize — practice explaining your answers, and apply these concepts in hands-on projects.
