AIDA 181 - Simulated Exam
A predictive model is applied to a clothing manufacturer's data of 1,000 employees, 50 of whom had workplace injuries in the past year. The table below shows how often the model correctly and incorrectly predicts for each employee "yes, will have an accident" or "no, will not have an accident." Predicted No+Predicted Yes=Total (1,000 Employees) Actual No 945 5 950 Actual Yes 10 40 50 Based on the preceding number, these statements can be made: There are 40 true positives (TP) for which the model correctly predicted yes. There are 945 true negatives (TN) for which the model correctly predicted no. There are 5 false positives (FP) for which the model incorrectly predicted yes (and the actual answer is no). There are 10 false negatives (FN) for which the model incorrectly predicted no (and the actual answer is yes). What is the recall of the workplace injury predictive model?
A. .80 B. .842 C. .889 D. .985 ** CORRECT: .80
If a predictive model makes 60 percent positive predictions in a situation in which without the model, only 40 percent of positive predictions would be made by chance, which one of the following is the model's lift?
A. 0.20 B. 0.67 C. 1.50 D. 2.00 ** CORRECT: 1.50
James, the risk manager for Paradox Construction (PC), suspected that some of its worker compensation claims involving back injuries were fraudulent. He suspected that some of these employees were actually injuring themselves over weekends. A text mining model was created to identify the word "weekend" in the interoffice email of the injured workers. A model was then created to see if there was a correlation between the use of the word "weekend" and fraudulent workers compensation claims. The exhibit below shows the predicted and actual results of the model. What is the precision of the text mining model? Out of 50 Emails "Weekend" present "Weekend" not present Not fraudulent 25 true negatives (TN) 5 false positives (FP)Fraudulent 5 false negatives (FN) 15 true positives (TP)
A. 0.30 B. 0.60 C. 0.75 D. 0.80 ** CORRECT: 0.75 precision = true positives / (true positives + false positives) precision = 15 / (15 + 5) precision = 0.75
Greatview Insurance developed a classification tree to predict whether an insured will renew its auto policy. Assume that the model classifies 600 instances from the training data as renewals based on the following attributes: homeowners coverage also with Greatview, customer tenure of more than three years, and a rate increase of less than 10%. However, there were actually 150 with these attributes in the training group that did not renew. What is the probability that classification produced by the leaf node with these attributes is correct?
A. 10% B. 25% C. 66% D. 75% ** CORRECT: 75%
Which one of the following statements regarding a classification tree is true?
A. A classification tree cannot be used to estimate probability. B. A classification tree can be used to estimate a numerical value. C. A classification tree can be used to estimate numerical values and class probability. D. A classification tree can be used to estimate class probability. ** CORRECT: A classification tree can be used to estimate class probability.
Encrypting data is an example of
A. A data governance program. B. A data security program. C. A regulatory compliance program. D. An enterprise risk management program ** CORRECT: A data security program.
Which one of the following defines the duties of a data steward?
A. A data steward measures data compliance. B. A data steward provides technological support. C. A data steward is an experienced business analyst. D. A data steward is a project manager. ** CORRECT: A data steward is an experienced business analyst.
Three Hills Insurance wants to know the ultimate value of auto liability claims for the calendar year that just ended. Which one of the following data modeling techniques is most appropriate?
A. A descriptive model B. A lift curve C. An unsupervised learning model D. A predictive model ** CORRECT: A predictive model
Usage-based insurance (UBI) is
A. A type of auto insurance in which the premium is based on the policyholder's driving behavior. B. Based on attributes such as driving record, driver education, years of driving experience, and credit score. C. A device that tracks driving habits regarding braking, acceleration, speed, cornering, lane shifting, left turns versus right turns, and the time of day the vehicle is driven. D. A risk-management technique used for fleet safety programs to monitor employees' driving behaviors but not to price insurance. ** CORRECT: A type of auto insurance in which the premium is based on the policyholder's driving behavior.
An advantage of deep learning over machine learning is its
A. Ability to work with incomplete data as well as performing new tasks like speech and image recognition. B. Performance of a specific calculation with the data it receives and then improving the accuracy of those results. C. Processes and algorithms that are less complex and generally easier to explain, helping with transparency. D. Analysis and gleaning of insights from large quantities of data to improve the accuracy of calculations over time. ** CORRECT: Ability to work with incomplete data as well as performing new tasks like speech and image recognition.
Goshen Insurance began offering usage-based insurance to personal auto policyholders. The assumption was that using telematics would allow Goshen to better segment drivers into appropriate rating classifications. Which one of the following is an example of a personal auto pricing attribute that would be generated through telematics, rather than a traditional attribute?
A. Acceleration B. Age of the driver C. Territory D. Use of the auto ** CORRECT: Acceleration
How can an insurer use data analytics to identify fraudulent claims?
A. An insurer could use predictive modeling to prioritize a claim for investigation based on the probability that it is fraudulent. Data mining techniques, such as text mining, social network analysis, and cluster analysis, can be used to extract the data. B. An insurer might leverage its data analytics tools and stored data on claims to help it make better business decisions based on past outcomes from the use of similar types of data. C. An insurer could tout its cutting-edge technology in material used to recruit talent, and it could enable its human resources to use predictive modeling techniques to develop the ability to analyze employee turnover and hiring patterns to identify claims recruits that will stay with the company. D. An insurer might use data analytics techniques to deploy information throughout the entire organization so that all staff members understand the value of intelligently deployed information and assist with data-driven decision making. ** CORRECT: An insurer could use predictive modeling to prioritize a claim for investigation based on the probability that it is fraudulent. Data mining techniques, such as text mining, social network analysis, and cluster analysis, can be used to extract the data.
Which one of the following best describes the purpose of applying a predictive model to holdout data during the training process?
A. Applying a predictive model to the holdout data helps ensure that the model will be cross-validated. B. Applying a predictive model to the holdout data helps ensure that the model is not overfitted to the training data. C. Applying a predictive model to the holdout data helps ensure that the model will not apply generalization, which is the ability of a model to apply itself to data outside the training data. D. Applying a predictive model to the holdout data helps ensure that the model will not become a confusion matrix. ** CORRECT: Applying a predictive model to the holdout data helps ensure that the model is not overfitted to the training data.
Donna's Dog Treats has been very successful in the Boston area and would like to expand to new cities. Donna knows that she cannot make this decision based on customer advice and blind faith. She has collected internal financial and operational data as well as external data from reliable sources. Donna has hired an analyst to review the data quality. The analyst is reviewing the data to see if it includes the demographics for each target city that Donna is considering. Which one of the following data-quality principles is being evaluated?
A. Appropriateness B. Validity C. Reasonableness D. Comprehensiveness ** CORRECT: Comprehensiveness
Through data mining, Goshen Mutual discovers that customers who insure two or more vehicles on a personal auto policy are very likely to buy a personal umbrella policy. Algorithms are then used to identify potential customers who might be interested in purchasing both personal auto and umbrella policies. Which one of the following data mining techniques did Goshen Mutual use?
A. Association rule learning B. Cluster analysis C. Classification D. Regression analysis ** CORRECT: Association rule learning
Wyatt Solutions' new risk manager uses data science to improve the organization's business results in all of the following ways, EXCEPT:
A. Automating decision making for greater accuracy B. Exploring new sources of data for predicting future losses C. Leveraging continued reliance on human analysis of data D. Organizing even larger amounts of new kinds of data ** CORRECT: Leveraging continued reliance on human analysis of data
A neural network can work with
A. Categorical values only. B. Numerical values only. C. Numerical and categorical values. D. Probability values only. ** CORRECT: Numerical values only.
Emilia is the workers compensation (WC) claims manager for Goshen Mutual. She is working on a project to better forecast the development of WC claims from first report to ultimate. Emilia has determined that back injury claims seem to be the most significantly underestimated, but she has no idea why. She has collected a database of 5 years of WC back injury claims, and requested the help of a data scientist to determine the reasons for the unusual development. Which one of the following unsupervised learning techniques will the data scientist most likely recommend in this situation?
A. Classification tree B. Correlation matrix C. Linear regression D. K-means cluster ** CORRECT: K-means cluster
Chloe is the personal auto claim supervisor for Parker Insurance. She notices an increase in bodily injury claims in one of her territories, and suspects that some of the claims may be fraudulent. Parker's internal fraud detection system identifies a number of claimants who were all being treated by the same medical group. Chloe suspects that a fraud ring may be involved. Which one of the following methods should Chloe use to help her identify the fraud ring?
A. Classification tree B. Linear Regression C. Network analysis D. Telematics ** CORRECT: Network analysis
Courtland Insurance uses various data science techniques to price personal auto insurance based on individual driver characteristics. Which one of the following techniques would most likely observe a driver characteristic from loss information that had not previously been considered as a rating factor?
A. Classification tree B. Linear regression C. Sentiment analysis D. Neural network ** CORRECT: Neural network
Which one of the following correctly describes cluster analysis?
A. Cluster analysis predicts a numerical value given characteristics of each member of a dataset. B. Cluster analysis assigns members of a dataset into categories based on known characteristics. C. Cluster analysis develops algorithms to develop rules to apply to new data. D. Cluster analysis explores data to find groups with common and previously unknown characteristics. ** CORRECT: Cluster analysis explores data to find groups with common and previously unknown characteristics.
Which one of the following functions of a data management program would allow accounting transactions to automatically update an organization's financial statements?
A. Data integration B. Data access C. Data governance D. Data preparation ** CORRECT: Data integration
Which one of the following provides the frame of reference needed so data can be used appropriately for analysis and decision-making?
A. Data virtualization B. Data lineage C. Data custodian D. Metadata ** CORRECT: Metadata
Tom, the regional manager at Westfork Mutual, has planned a community service day for all employees. He has asked his two assistant managers, Julian and Leah, to spread the word to other employees and encourage them to attend. Based on centrality measures of the company's email traffic, Tom is confident that Julian will spread the word to more employees more quickly through email than will Leah. Tom's confidence is based on Julian's high score in which one of the following centrality measures?
A. Degree B. Betweenness C. Closeness D. Association ** CORRECT: Closeness
A type of traditional analysis allows a risk manager to examine the series of events and conditions that led to an accident; it interrupts the sequence of events leading to system failure so that the failure itself can be prevented. This type of analysis is called
A. Failure mode and effect analysis. B. Criticality analysis. C. Root cause analysis. D. Fault tree analysis. ** CORRECT: Fault tree analysis.
Durham Insurance (DI) provides personal lines coverage to policyholders throughout the U.S. A local newspaper in New Jersey has accused DI of unfair and deceptive claims practices following a recent hurricane. The risk manager works with data scientists to mine news and social media posts since the release of the initial news article. They learn that the negative sentiment has spread in a short period of time and that they must implement an immediate response to defend its reputation. Which one of the following tools would help DI determine which stakeholder to focus on, and the most influential places to post responses?
A. Fault tree B. Sociogram C. Classification tree D. Process map ** CORRECT: Sociogram
Which one of the following is true regarding underwriting products liability risks using data mining?
A. For assessing liability risks for an emerging technology, a predictive model may not be useful as a first step. B. In this insurance effort, the use of text mining is limited to social media. C. Predictive modeling requires known attributes but not a known target variable. D. A predictive model is an appropriate tool to be used prior to cluster analysis. ** CORRECT: For assessing liability risks for an emerging technology, a predictive model may not be useful as a first step.
Millstone Insurance wants to use the concept of similarity and distance to help identify profitable prospects in its target market of small restaurants. The underwriters have identified a current policyholder (CPH) that they consider a desirable risk, and would like the data scientist to help create a data mining technique to identify similar restaurants. Which one of the following is the first step in using distance to measure similarity?
A. Identify which prospects are the nearest neighbors to CPH B. Make a list of all of the common attributes between CPH and the prospective restaurants C. Decide which of CPH's attributes indicate a desirable risk D. Plot all of the prospective restaurants on a graph based on geographic distance from CPH ** CORRECT: Decide which of CPH's attributes indicate a desirable risk
Alignment of the data analytics project's goals with upper management's long-term organizational objectives is an example of
A. Identifying risk and its impact on the data analytics project. B. Component activities associated with the data analytics initiative. C. A type of data mining technique that yields practical results. D. One method of treating risks that have been analyzed. ** CORRECT: Component activities associated with the data analytics initiative.
Which one of the following data governance tools allows the data governance committee to look at data relationships and interdependencies across the organization?
A. Internal coding procedures B. Enterprise data models C. Project management programs D. External compliance guidelines ** CORRECT: Enterprise data models
The process map in the application of process mining to claims handling
A. Is a predictive model that allows the insurer to understand how long each new claim will take to conclude. B. Allows the insurer to understand the average length of time for each process instance in the current claims process. C. Diagrams methods in which the insurer could improve the time required for each process instance in the claims process. D. Shows how the entire claims process could be improved by applying data analytics to process discovery. ** CORRECT: Allows the insurer to understand the average length of time for each process instance in the current claims process.
Which one of the following is an advantage of using radio frequency identification (RFID) for supply chain management?
A. It can capture images that aid in loss prevention and reduction. B. It identifies assets and compiles their characteristics without human intervention. C. It can monitor illumination and provide lighting when needed. D. It can assess impending fire conditions so that first responders can be notified before a fire begins. ** CORRECT: It identifies assets and compiles their characteristics without human intervention.
Jaime is an experienced claims representative for Rely Insurance and is using traditional fraud indicators as well as mining social media data to help identify fraudulent claims assigned to her department. Regarding this, which one of the following is true?
A. Jaime discovers fraud throughout her twenty-year career has pretty much remained the same. B. Searching her insureds' and claimants' social media posts is Jaime's thorough attempt to stay in front of potential fraudsters. C. Jamie searches Rely's list of fraud indicators as well as consulting those from fraud-fighting organizations as she begins her work. D. Network analysis and predictive modeling work well in isolation but are conflicting methods when used together. ** CORRECT: Jamie searches Rely's list of fraud indicators as well as consulting those from fraud-fighting organizations as she begins her work.
Facial recognition software has potential uses in identifying criminals and airport security. The emerging technologies used in facial recognition software are
A. Lidar and artificial intelligence. B. Wireless sensor networks and drones. C. Wearables and wireless sensor networks. D. Computer vision and artificial intelligence. ** CORRECT: Computer vision and artificial intelligence.
Class probability estimation is accomplished through
A. Linear regression. B. Support vector machines. C. Logistic regression. D. Linear discriminants. ** CORRECT: Logistic regression.
The efficiency of the flow between social network connections can be determined through
A. Local variables. B. Homophily. C. Centrality measures. D. Logistic regression. ** CORRECT: Centrality measures.
Wyatt Insurers has used robotic process automation enhanced with another technology that allows its computers to continually teach themselves to make better decisions based on previous results and new data. This other technology is
A. Machine learning. B. Computer vision. C. Natural language processing. D. Blockchain. ** CORRECT: Machine learning.
To effectively articulate throughout the organization the need for a shift to data-driven processes as part of a data analytics initiative, top management should
A. Mandate that the entire organization will adopt data-driven processes because management has determined that such an approach would be the best way to ensure the success of the organization in the long run. B. Require that employees embrace data-driven processes and learn to use data analytics or else they will be terminated, because management has determined that to be the best strategy to ensure the success of the organization. C. Arrange for training for all employees to learn to use data-driven processes and convince staff that they are preferred and then later inform staff that the organization is adopting such processes. D. Frame it as being urgent and crucial to the insurer's ability to acquire and retain business because competitors use data analytics techniques to improve their underwriting, pricing, fraud detection, and marketing. ** CORRECT: Frame it as being urgent and crucial to the insurer's ability to acquire and retain business because competitors use data analytics techniques to improve their underwriting, pricing, fraud detection, and marketing.
Which one of the following is an algorithm often used in cluster analysis?
A. Nearest neighbor B. Sentiment score C. K-means D. Centroid ** CORRECT: K-means
Which one of the following pairs represents internal environmental factors that could be affected by a data analytics initiative?
A. Opportunities and strengths B. Strengths and weaknesses C. Threats and opportunities D. Weaknesses and threats ** CORRECT: Strengths and weaknesses
The method whereby a computer will continually refine a model as additional data and results are received is known as
A. Progressive analysis. B. Machine learning. C. Regressive analysis. D. Overfitting ** CORRECT: Machine learning.
Which one of the following types of data is used to test a predictive model?
A. Recursive data B. Training data C. Holdout data D. Lift data ** CORRECT: Holdout data
Under the General Data Protection Regulation (GDPR), a data controller's role is to
A. Represent the business aspects of data governance. B. Manage the flow of data for the rest of the organization. C. Define how and for what purpose personal data should be processed. D. Define the metrics used to measure an organization's overall data quality. ** CORRECT: Define how and for what purpose personal data should be processed.
An insurer wants to determine the complexity of a claim at the time of first report. Which one of the following is an important step after identifying the attributes of complex claims?
A. Selecting one or two of the attributes to develop a predictive model B. Developing a classification tree with each attribute having equal importance C. Ranking the attributes according to their relative information gain D. Using each of the attributes to form a leaf node in a classification tree ** CORRECT: Ranking the attributes according to their relative information gain
In order to be useful, data should be
A. Structured. B. Unstructured. C. Structured or unstructured. D. Internal. ** CORRECT: Structured or unstructured.
Greatview Insurance is a regional insurer. The profitability of its homeowners book has been declining for several years. In the past, Greatview has been able to maintain profitability by using traditional rating variables, and implementing flat rate increases across the entire book when necessary. Management believes that is the only way to improve profitability. Which one of the following would most likely occur if Greatview continues to apply flat rate increases across the homeowners book?
A. The customers with the lower loss ratios would most likely find coverage elsewhere, and the overall retention and loss ratio would increase. B. The customers with higher loss ratios would be more likely to stay, leaving the insurer with the least profitable segment of its book. The overall loss ratio would most likely not improve. C. The customers with higher loss ratios would most likely find coverage elsewhere, and the overall retention and loss ratio would decrease. D. The customers with lower loss ratios would most likely pay the increase and stay, and the customers with the higher loss ratio would leave, resulting in an improved loss ratio for the book. ** CORRECT: The customers with higher loss ratios would be more likely to stay, leaving the insurer with the least profitable segment of its book. The overall loss ratio would most likely not improve.
If clusters are grouped at one level of the hierarchy,
A. They are only grouped at that one level of the hierarchy. B. They are grouped at all levels of the hierarchy. C. They are grouped at each lower level of the hierarchy. D. They are grouped at each higher level of the hierarchy. ** CORRECT: They are grouped at each higher level of the hierarchy.
The most important reason for risk management and insurance professionals to learn about data analytics is
A. To be able to conduct analyses of social media. B. To design analytical models in their areas of expertise. C. Because big data and technology are central to the insurance industry. D. To propose new ideas for technology to provide data. ** CORRECT: Because big data and technology are central to the insurance industry.
Westfolk Mutual has formed a team to analyze why its retention of homeowners policies is declining. Traditional data analysis indicates that many policyholders are moving to a new insurer after three years with Westfolk Mutual, but there is no clear reason for the move. Which one of the following best describes the role of the team's data scientist?
A. To predict the retention rate of homeowners policies for the current year B. To determine the qualities of policyholders that should be retained C. To provide the domain knowledge for the team D. To explore social media, text, and geolocation-based data for reasons ** CORRECT: To explore social media, text, and geolocation-based data for reasons
Which one of the following best describes the function of the service layer of a smart operation?
A. Transmitting data using wireless protocols that ensure necessary capabilities B. Transmitting data without human interaction C. Employing applications that use data processing, cloud computing, and storage and analysis of large amounts of data D. Using sensors, cameras, and data-collection capabilities to provide information to others ** CORRECT: Employing applications that use data processing, cloud computing, and storage and analysis of large amounts of data
