IS4242 Quiz 2 (Practice) PDF 2024/2025
Document Details
Uploaded by DexterousFern6890
NUS
2024
null
Tags
Summary
This document is a 2024/2025 sample quiz from the IS4242 Intelligent Systems and Techniques course. It contains questions about proactive and reactive churn management, neural networks, and other relevant topics.
Full Transcript
IS4242— Intelligent Systems and Techniques Quiz - 2 (Practice) 2024/2025 Semester 1 Time allowed: 4:00 - 5:15 P.M. (1 hour 15 min) Date: 12 Nov 2024 Instructions (please read car...
IS4242— Intelligent Systems and Techniques Quiz - 2 (Practice) 2024/2025 Semester 1 Time allowed: 4:00 - 5:15 P.M. (1 hour 15 min) Date: 12 Nov 2024 Instructions (please read carefully): STUDENT NUMBER A 1. Write down your student number on the right us- U 0 0 0 0 0 0 0 A N ing ink or pencil, and shade the corresponding cir- A 1 1 1 1 1 1 1 B R cle in the grid for each digit or letter. DO NOT HT 2 2 2 2 2 2 2 E U WRITE YOUR NAME! NT 3 3 3 3 3 3 3 H W 4 4 4 4 4 4 4 J X 2. This answer booklet comprises of 8 pages, includ- 5 5 5 5 5 5 5 L Y 6 6 6 6 6 6 6 M ing this cover page. 7 7 7 7 7 7 7 8 8 8 8 8 8 8 3. There are 24 questions in total, each question car- 9 9 9 9 9 9 9 ries 1 mark. There is no negative marking for wrong answers. The maximum points you can score is 24. 5cm For Examiner’s Use Only 4. All questions must be answered in the space pro- vided; no extra sheets will be accepted as answers. Page Points Score 5. You must submit this entire answer booklet. 3 5 4 5 6. This is a CLOSED BOOK assessment. You are not allowed to bring in any printed materials. 5 5 7. Calculators are allowed, but laptops or other elec- 6 5 tronic devices are not. 7 4 8. Leave your student card on the desk throughout Total: 24 this assessment. 9. DO NOT look at the questions until you are told to do so. 10. You are allowed to use pencils or ball pens as you like as long as it is legible (no red color, please). 11. Answer all the questions using the knowledge you learned in IS4242. This page is intentionally left blank. Page 2 1 MCQ Shade the bubbles to mark your choice. All questions have only one answer, select the best suitable choice. 1. (1 mark) In customer churn management, which of the following is NOT an example of a proactive churn management strategy? √ Sending win-back offers to lost customers ⃝ Monitoring customer satisfaction through surveys ⃝ Providing excellent customer service ⃝ Reacting to customer complaints 2. (1 mark) Which of the following factors may contribute to voluntary churn? ⃝ Effective loyalty programs ⃝ High switching costs √ Customer dissatisfaction ⃝ Excellent customer service 3. (1 mark) Which of the following is considered a reactive churn management strategy? ⃝ Identifying high-value customers for special offers √ Offering discounts to customers considering leaving ⃝ Conducting customer satisfaction surveys ⃝ Implementing a comprehensive loyalty program 4. (1 mark) A company uses predictive analytics to identify customers at high risk of churn. What is this strategy called? √ Proactive churn management ⃝ Reactive churn management ⃝ Involuntary churn ⃝ None of the above 5. (1 mark) What is the main objective of a Customer Churn Prediction model? ⃝ Identifying which customers are already lost. √ Predicting the likelihood of future customer churn. ⃝ Estimating the revenue generated by loyal customers. ⃝ Determining the optimal pricing strategy. Page 3 6. (1 mark) In the context of training neural networks, how do backpropagation and activation functions interact with each other? ⃝ Backpropagation determines the number of hidden layers, while activation functions optimize the weights. ⃝ Backpropagation calculates the loss function, while activation functions adjust the learning rate. √ Backpropagation computes the gradient of the loss with respect to the net- work’s weights, and activation functions introduce non-linearity to the net- work. ⃝ Backpropagation selects the optimization algorithm, while activation functions con- trol the regularization terms. 7. (1 mark) What is the contemporary definition of a brand in the context of marketing? ⃝ A company’s logo and slogan. ⃝ The physical packaging of a product. √ The perception and image that consumers have about a product or company. ⃝ The price of a product in the market. 8. (1 mark) What is an essential aspect of creating and managing a brand successfully? ⃝ Changing the brand identity frequently to keep it fresh. √ Consistently delivering on the brand promise to customers. ⃝ Copying the branding strategies of successful competitors. ⃝ Focusing only on short-term sales without considering the long-term brand image. 9. (1 mark) How can a firm measure brand image among consumers? ⃝ By examining the company’s financial statements. ⃝ By analyzing the number of products sold. √ By analyzing customer surveys and feedback. ⃝ By assessing the company’s employee satisfaction. 10. (1 mark) Which of the following is an example of co-branding? ⃝ A brand changing its logo to make it look similar to competitors. √ An apparel brand and a footwear brand form an alliance to offer exclusive products. ⃝ A company expanding its product line by adding new features to existing products. ⃝ A brand increasing its advertising budget for the existing products. Page 4 11. (1 mark) In the context of CNNs, what is the purpose of applying padding to the input (e.g., picture or feature map of the picture) before conducting convolution operations? ⃝ Padding increases the spatial resolution of the feature map for more detailed feature extraction. ⃝ Padding reduces the computational complexity of the convolutional operation. √ Padding helps prevent the loss of information at the edges of the feature map. ⃝ Padding introduces non-linear activation functions to enhance feature representa- tion. 12. (1 mark) Consider a scenario where grid search is used for hyperparameter tuning, and the search space includes continuous hyperparameters. What is an appropriate strategy to handle continuous hyperparameter values within a grid search? √ Pick a few values at fixed intervals. ⃝ Exclude continuous hyperparameters from the grid search. ⃝ Perform a random search for continuous hyperparameters separately. ⃝ None of the above. 13. (1 mark) Lazada has started an initiative to give a promotion to customers who give effective feedback to their purchased products. The initiative coincides with the generation of: ⃝ Network effects √ Electronic Word of Mouth ⃝ Voluntary churn ⃝ Switching costs 14. (1 mark) In order to safeguard the privacy of the reviewers, Amazon intends to implement an anonymous reviewer system. How does the lack of transparency on reviewer identity impact the content and the credibility of customer feedback? ⃝ Implementing stricter regulations on reviewer anonymity would hinder the diversity of opinions and experiences shared on the platform. ⃝ Transparency in reviewer identity has no significant impact on the perceived credi- bility of reviews on Amazon. √ The absence of reviewer identity may compromise the reliability of feedback, posing challenges in distinguishing genuine reviews from potential manipula- tion. ⃝ None of the above 15. (1 mark) Which of the following is the negative side of Electronic Word of Mouth (eWOM)? √ Fake news spreads much faster. ⃝ People are naturally convinced by the feedback from their family and friends. ⃝ eWOM is mostly non-commercial. Page 5 ⃝ People are often driven by economic incentive. 16. (1 mark) What psychological factors often motivate consumers to share their opinions, expe- riences, or reviews about a product or brand online? ⃝ A desire for anonymity and privacy in online interactions. √ Intrinsic motivations such as the feeling of pleasure obtained from helping oth- ers. ⃝ Fear of social backlash and criticism from online communities. ⃝ None of the above 17. (1 mark) Which of the following is NOT true for Word2Vec representation of words? ⃝ Capture the meaning of a word in terms of neighbouring words. ⃝ A neural network with a single hidden layer is enough to obtain Wrod2Vec repre- sentations. √ Training data preparation requires a lot of human effort. ⃝ Outputs lower dimensional representations compared to term frequency - inverse document frequency (tf-idf) representation. 18. (1 mark) Bidirectional Long Short Term Memory (BiLSTM) networks are favored for their ability to capture context and dependencies in sequences. How does the bidirectional pro- cessing of BiLSTM enhance its performance compared to unidirectional LSTMs for tasks like sentiment analysis? ⃝ BiLSTM uses quite a large (> 2) number of hidden layers to process sequences more efficiently than unidirectional LSTMs. ⃝ BiLSTM can process sequences in parallel, while unidirectional LSTMs process sequences sequentially. √ BiLSTM captures information in both forward and backward directions, pro- viding a more comprehensive understanding of the sequence. ⃝ BiLSTM utilizes different activation functions compared to unidirectional LSTMs for information processing. 19. (1 mark) Sustainable competitive advantage in business is achieved when a company: ⃝ Focuses on cost-cutting measures. √ Develops unique capabilities that are difficult to imitate. ⃝ Increases product prices to maximize profits. ⃝ Maintains a strategy and ignores the changes in the market environment. 20. (1 mark) What does the term “Value Chain” refer to in strategic management? ⃝ A chain of command within an organization √ A sequence of activities that add value to a product or service Page 6 ⃝ The company’s marketing strategy ⃝ The link between a company’s mission and vision 21. (1 mark) When considering the value chain, which of the following represents the primary activity that directly adds value to a product or service? ⃝ Customer Support ⃝ Marketing and Sales √ Operations ⃝ Outbound Logistics 22. (1 mark) In the context of Porter’s Five Forces, which of the following factors typically rep- resents a key determinant of the intensity of rivalry among existing competitors in an industry, potentially making it more challenging for firms to achieve sustainable profitability? √ High product quality and differentiation ⃝ A high barrier to entry for new firms ⃝ A relatively small number of suppliers ⃝ None of the above 23. (1 mark) What are the unique advantages obtained by using graph or network analysis to understand business relationships? ⃝ Graph analysis is irrelevant in business contexts as simple statistical methods (his- tograms, summary statistics, correlation analysis, etc.) provide sufficient insights. √ Graph analysis enhances the visualization of complex relationships, facilitating a deeper understanding of network structures within and across organizations. ⃝ Utilizing graph analysis in business is primarily focused on quantifying financial metrics rather than relationship dynamics. ⃝ Graph analysis is suited for analyzing relationship dynamics among humans but not for organizations. 24. (1 mark) When applying the Louvain community detection algorithm over a network (With a large number of nodes and edges), how does the algorithm optimize modularity? ⃝ The Louvain algorithm maximizes modularity by partitioning the network into com- munities based on the random assignment of nodes. √ The Louvain algorithm iteratively moves nodes between communities to maxi- mize the network’s modularity. ⃝ The Louvain algorithm maximizes network’s modularity by merging all nodes at once into a single community. ⃝ Modularity optimization in the Louvain algorithm is irrelevant to community detec- tion, as it focuses solely on network structure. Page 7 – END OF PAPER – Page 8