Q1. What is typically the biggest challenge in assessing credit risk?
- A. High-income applicants
- B. People with the best credit don't need loans, and those with the worst likely won’t repay
- C. Bank staff being biased
-
D. Low interest rates
Answer: B. People with the best credit don't need loans, and those with the worst likely won’t repay
Q2. Which of the following is considered when evaluating someone's credit history?
- A. Tax returns
- B. Loan type
- C. Whether previous loans were paid on time
-
D. Age of the applicant
Answer: C. Whether previous loans were paid on time
Q3. Why are middle-segment customers typically preferred by banks?
- A. They demand higher interest
- B. They are more likely to need loans and repay them
- C. They have government backing
-
D. They always default
Answer: B. They are more likely to need loans and repay them
Q4. What attribute does NOT typically influence credit risk?
- A. Marital status
- B. Annual income
- C. Favorite color
-
D. Credit history
Answer: C. Favorite color
Q5. Which of these is a common loan term parameter?
- A. Type of currency
- B. Duration of loan repayment
- C. Loan repayment office
-
D. Weekly groceries
Answer: B. Duration of loan repayment
Q6. Income of a loan applicant is an example of a:
- A. Dependent variable
- B. Noise variable
- C. Independent variable
-
D. Categorical label
Answer: C. Independent variable
Q7. A married couple applying for a home loan provides which kind of data?
- A. Loan default record
- B. Personal information
- C. Output variable
-
D. Noise factor
Answer: B. Personal information
Q8. Credit risk models are typically built using:
- A. Surveys
- B. Machine learning
- C. Manual inspection
-
D. Stock market trends
Answer: B. Machine learning
Q9. Successful credit risk modeling has led to:
- A. Decrease in credit card usage
- B. Proliferation of mortgages and credit cards
- C. Higher default rates
-
D. Less bank automation
Answer: B. Proliferation of mortgages and credit cards
Q10. A bank rejecting a loan because of poor credit score is a:
- A. Random process
- B. Classification decision
- C. Regression decision
-
D. Time series output
Answer: B. Classification decision
Q11. Credit risk classification is an example of:
- A. Supervised learning
- B. Unsupervised learning
- C. Clustering
-
D. Reinforcement learning
Answer: A. Supervised learning
Q12. Which of the following is the class label in credit risk classification?
- A. Annual income
- B. Home ownership
- C. Loan default (Yes/No)
-
D. Marital status
Answer: C. Loan default (Yes/No)
Q13. In a training set, what is meant by x?
- A. Class label
- B. Outcome
- C. Attributes or input variables
-
D. Test case
Answer: C. Attributes or input variables
Q14. What type of classifier technique uses similarity to known points?
- A. Naïve Bayes
- B. Decision Tree
- C. Nearest-Neighbor
-
D. Logistic Regression
Answer: C. Nearest-Neighbor
Q15. Which classifier method builds hierarchical tree-like structures?
- A. Neural networks
- B. K-means
- C. Decision Trees
-
D. SVM
Answer: C. Decision Trees
Q16. Which ensemble method uses multiple trees?
- A. Naïve Bayes
- B. Random Forest
- C. KNN
-
D. SVM
Answer: B. Random Forest
Q17. Why are decision trees easy to interpret?
- A. They use linear equations
- B. They model logic rules
- C. They cluster data automatically
-
D. They need no training data
Answer: B. They model logic rules
Q18. What is a necessary requirement for a decision tree model?
- A. No class labels
- B. Image data
- C. Predefined discrete classes
-
D. Only numeric input
Answer: C. Predefined discrete classes
Q19. Which of the following would most likely result in a loan default?
- A. Income over $150K
- B. Married and homeowner
- C. No home, low income, single
-
D. Good credit history
Answer: C. No home, low income, single
Q20. A model that learns from labeled examples is called:
- A. Unsupervised
- B. Reinforced
- C. Supervised
-
D. Clustering
Answer: C. Supervised
Q21. What does a decision tree split data based on?
- A. Correlations
- B. Random noise
- C. Attribute values
-
D. Time
Answer: C. Attribute values
Q22. In a decision tree, leaves represent:
- A. Feature sets
- B. Nodes
- C. Final decisions
-
D. Regression equations
Answer: C. Final decisions
Q23. Decision trees are useful because they:
- A. Require no computation
- B. Can represent rules
- C. Always yield 100% accuracy
-
D. Are slower than neural nets
Answer: B. Can represent rules
Q24. If two trees fit the same data, this is because:
- A. Data is corrupted
- B. Classification is not needed
- C. Multiple rules can produce same results
-
D. Machine learning failed
Answer: C. Multiple rules can produce same results
Q25. Classification assumes:
- A. No training data
- B. Continuous outcome variable
- C. Predefined class labels
-
D. Supervised clustering
Answer: C. Predefined class labels
Q26. A person who owns a home and earns >80K is:
- A. Likely to default
- B. Likely not to default
- C. Indeterminate
-
D. Not a loan candidate
Answer: B. Likely not to default
Q27. If a person is single, not a homeowner, and earns <80K, the decision tree would likely:
- A. Approve the loan
- B. Request more documents
- C. Predict default
-
D. Ask for a co-signer
Answer: C. Predict default
Q28. Marital status and home ownership are examples of:
- A. Output labels
- B. Class variables
- C. Input attributes
-
D. Noise features
Answer: C. Input attributes
Q29. What is used to test a decision tree after it is trained?
- A. Training data
- B. Evaluation metrics
- C. Test set
-
D. Parameters
Answer: C. Test set
Q30. What is the process of making predictions called?
- A. Induction
- B. Deduction
- C. Regression
-
D. Forecasting
Answer: B. Deduction
Q31. Which of the following is NOT typically used in a decision tree?
- A. Root node
- B. Leaf node
- C. Branch
-
D. Circle loop
Answer: D. Circle loop
Q32. A classification rule such as “If A and B, then Yes” is represented in:
- A. Linear Regression
- B. Naïve Bayes
- C. Decision Tree
-
D. Clustering
Answer: C. Decision Tree
Q33. The process of using training data to build a model is called:
- A. Pruning
- B. Testing
- C. Training
-
D. Validating
Answer: C. Training
Q34. Which classification technique is most sensitive to outliers?
- A. Random Forest
- B. Decision Trees
- C. K-Nearest Neighbors
-
D. SVM
Answer: C. K-Nearest Neighbors
Q35. Which of the following is true about overfitting?
- A. The model generalizes well
- B. It performs poorly on training data
- C. It performs well on training but poorly on new data
-
D. It increases prediction accuracy
Answer: C. It performs well on training but poorly on new data
Q36. Pruning in decision trees helps to:
- A. Increase tree depth
- B. Improve training accuracy
- C. Reduce overfitting
-
D. Add more features
Answer: C. Reduce overfitting
Q37. Which metric would you use to assess classification performance?
- A. R-squared
- B. Mean Absolute Error
- C. Accuracy
-
D. Median
Answer: C. Accuracy
Q38. In credit modeling, a confusion matrix helps analyze:
- A. Decision speed
- B. Loan tenure
- C. Model performance
-
D. Customer feedback
Answer: C. Model performance
Q39. A model that has high precision but low recall:
- A. Returns many false positives
- B. Misses many actual positives
- C. Is ideal for all cases
-
D. Overfits always
Answer: B. Misses many actual positives
Q40. Decision trees handle:
- A. Only numeric data
- B. Only categorical data
- C. Both numeric and categorical data
-
D. Only binary output
Answer: C. Both numeric and categorical data
Q51. Customer churn refers to:
- A. Customers buying more
- B. Customers canceling services or subscriptions
- C. Returning customers
-
D. Product returns
Answer: B. Customers canceling services or subscriptions
Q52. Churn is a key concern in:
- A. Manufacturing
- B. Telecom and banking
- C. Construction
-
D. Government
Answer: B. Telecom and banking
Q53. What kind of machine learning is churn prediction?
- A. Unsupervised learning
- B. Supervised learning
- C. Reinforcement learning
-
D. Dimensionality reduction
Answer: B. Supervised learning
Q54. In churn prediction, the class label is usually:
- A. Type of service
- B. Subscription duration
- C. Churn (Yes/No)
-
D. Age group
Answer: C. Churn (Yes/No)
Q55. A churn prediction model can help businesses to:
- A. Expand manufacturing
- B. Improve recruitment
- C. Retain customers
-
D. Lower taxes
Answer: C. Retain customers
Q56. Which of these features is least useful in churn prediction?
- A. Customer complaints
- B. Tenure with company
- C. Monthly charges
-
D. Customer’s height
Answer: D. Customer’s height
Q57. High monthly charges may indicate:
- A. Happy customer
- B. Potential churn
- C. No correlation
-
D. Data error
Answer: B. Potential churn
Q58. Churn prediction models must be:
- A. Perfectly accurate
- B. Interpretable and timely
- C. Expensive and slow
-
D. Written in Java
Answer: B. Interpretable and timely
Q59. An unbalanced dataset for churn prediction may have:
- A. Equal churn and no-churn
- B. Mostly churn cases
- C. Mostly non-churn cases
-
D. Only new customers
Answer: C. Mostly non-churn cases
Q60. In churn prediction, recall is more important than:
- A. F1-score
- B. Precision (in some cases)
- C. Support
-
D. Accuracy
Answer: B. Precision (in some cases)
Q61. Upselling to high-risk churn customers is:
- A. Profitable strategy
- B. Bad business logic
- C. Churn mitigation
-
D. Data cleansing
Answer: C. Churn mitigation
Q62. A false negative in churn prediction means:
- A. Predicting churn when it’s not
- B. Missing an actual churn
- C. Predicting loyalty correctly
-
D. None of the above
Answer: B. Missing an actual churn
Q63. The goal of churn prediction is to:
- A. Hire better agents
- B. Know which customers to offer retention deals to
- C. Reduce server load
-
D. Create ads
Answer: B. Know which customers to offer retention deals to
Q64. In logistic regression used for churn, the output is:
- A. A class label directly
- B. A probability between 0 and 1
- C. A decision tree
-
D. A clustering group
Answer: B. A probability between 0 and 1
Q65. The feature “Tenure < 1 year” might signal:
- A. Loyal customer
- B. High risk of churn
- C. No relation
-
D. Data error
Answer: B. High risk of churn
Q66. A high churn rate indicates:
- A. High retention
- B. Many new customers
- C. Many lost customers
-
D. None of the above
Answer: C. Many lost customers
Q67. Which action can reduce churn?
- A. Ignoring complaints
- B. Offering long-term discounts
- C. Reducing support team
-
D. Increasing fees
Answer: B. Offering long-term discounts
Q68. Which of the following is a retention strategy?
- A. Cancelling all low-usage users
- B. Sending loyalty rewards
- C. Reducing data collection
-
D. Lowering response times
Answer: B. Sending loyalty rewards
Q69. Segmenting customers by churn risk helps:
- A. Reduce ads
- B. Prioritize outreach
- C. Increase complaints
-
D. Decrease accuracy
Answer: B. Prioritize outreach
Q70. Decision tree models for churn are valued because:
- A. They predict probabilities
- B. They model time series
- C. They provide simple if-then rules
-
D. They cluster customers
Answer: C. They provide simple if-then rules
Q81. KNIME is primarily used for:
- A. Video editing
- B. Data analysis and machine learning
- C. Web development
-
D. Drawing diagrams
Answer: B. Data analysis and machine learning
Q82. KNIME stands for:
- A. Knowledge Integration Mining Environment
- B. Known Intelligent Machine Evaluator
- C. Kernel in Machine Exploration
-
D. Knowledge Network Infrastructure for Modelling Experiments
Answer: A. Knowledge Integration Mining Environment
Q83. In KNIME, what is a “workflow”?
- A. A document file
- B. A series of connected nodes
- C. An Excel macro
-
D. A Python script
Answer: B. A series of connected nodes
Q84. Which file type is commonly used to import data in KNIME?
- A. .mp3
- B. .xml
- C. .csv
-
D. .exe
Answer: C. .csv
Q85. KNIME nodes are:
- A. Scripts
- B. Plugins
- C. Visual blocks with specific tasks
-
D. Databases
Answer: C. Visual blocks with specific tasks
Q86. KNIME’s visual interface allows for:
- A. Text-only inputs
- B. Manual script writing only
- C. Drag-and-drop modeling
-
D. GPU rendering
Answer: C. Drag-and-drop modeling
Q87. What kind of learning does KNIME support?
- A. Supervised only
- B. Unsupervised only
- C. Both supervised and unsupervised
-
D. None
Answer: C. Both supervised and unsupervised
Q88. KNIME integrates with which programming language?
- A. Java
- B. Python
- C. R
-
D. All of the above
Answer: D. All of the above
Q89. KNIME workflows can be saved and:
- A. Shared and reused
- B. Only run once
- C. Sent via email as MP3
-
D. Uploaded to Instagram
Answer: A. Shared and reused
Q90. KNIME excels in:
- A. High-end graphics
- B. Text editing
- C. Visual programming for data science
-
D. Backend development
Answer: C. Visual programming for data science
Q91. To build a decision tree in KNIME, you must:
- A. Write Java code
- B. Connect a decision tree learner and predictor node
- C. Train a deep learning model
-
D. Use Excel macros
Answer: B. Connect a decision tree learner and predictor node
Q92. KNIME’s output nodes are used to:
- A. Hide the data
- B. Export or visualize results
- C. Encrypt files
-
D. Delete logs
Answer: B. Export or visualize results
Q93. KNIME supports integration with:
- A. Azure
- B. TensorFlow
- C. H2O.ai
-
D. All of the above
Answer: D. All of the above
Q94. What makes KNIME attractive for non-programmers?
- A. Free food coupons
- B. Command-line only tools
- C. Visual, no-code workflow design
-
D. Mandatory Java knowledge
Answer: C. Visual, no-code workflow design
Q95. You can automate workflows in KNIME by:
- A. Repeating steps manually
- B. Using workflow loops and scheduling
- C. Installing more RAM
-
D. None of the above
Answer: B. Using workflow loops and scheduling
Q96. Which node would you use to read an Excel file in KNIME?
- A. Java Reader
- B. Table Reader
- C. Excel Reader
-
D. Data Generator
Answer: C. Excel Reader
Q97. KNIME's key advantage over coding tools is:
- A. Data loss
- B. Visual drag-and-drop interface
- C. More bugs
-
D. Poor compatibility
Answer: B. Visual drag-and-drop interface
Q98. KNIME Server adds functionality such as:
- A. No extra features
- B. Scheduled workflow execution and collaboration
- C. Video editing tools
-
D. Animation libraries
Answer: B. Scheduled workflow execution and collaboration
Q99. KNIME offers:
- A. Paid-only plans
- B. A free open-source desktop version
- C. Only cloud-based usage
-
D. Subscription-based email
Answer: B. A free open-source desktop version
Q100. KNIME is best suited for:
- A. Video streaming
- B. Office automation
- C. Data science and predictive analytics
-
D. Entertainment
Answer: C. Data science and predictive analytics
