ALU 202 ACTUAL STUDY GUIDE QUESTIONS AND
ANSWERS SET A+
✔✔In triage underwriting - ✔✔rather than obtaining an underwriting requirement
because it is indicated on an age-and-amount chart, the score from a predictive model
is used to determine which underwriting requirements are needed, including whether
certain requirements are needed at all.
✔✔Propensity Scoring - ✔✔where cases have more detailed info collected, such as lab
values, then more complex models can be built to estimate the likelihood or propensity
of an insured having a certain condition or disease.
✔✔Benefits of Predictive Modeling in Underwriting - ✔✔Improved mortality and more
competitive pricing
Faster case processing and less invasive underwriting
Lower underwriting costs
Better underwriter utilization
✔✔Challenges with Predictive Modeling for Underwriting - ✔✔Data availability
Data quality
Model fitting and subject matter expertise
✔✔A build dataset - ✔✔that is used to develop the model
✔✔A validation dataset - ✔✔that is used to test the model
✔✔Blind spots - ✔✔involve a model that is not predictive for certain age banks or other
segments of the population since the model had no basis in its build data set.
✔✔Linear Regression Model - ✔✔may be the most frequently used type of model to
represent business process.
, ✔✔Logistic Regression - ✔✔is similar to a linear regression model with the important
difference being that the target variable is binary (yes/no) in nature.
✔✔Cox proportional hazard model - ✔✔is the most widely used statistical technique for
estimating individual risk in studies of survival.
✔✔Decision Tree Model - ✔✔divides a dataset into progressively smaller sub-
segments, using the features in rules.
✔✔Decision Tree Model Advantage - ✔✔It has the advantage of being simple to
understand and interpret, however, it may not generalize well from the training data,
resulting in overfitting.
✔✔Random Forests - ✔✔Use multiple decision trees based on different subsets of the
data and/or different subsets of features, and outputs the mode or mean prediction of
the individual tree.
✔✔Random Forests Benefits - ✔✔This method helps to reduce the decision trees'
limitation of overfitting.
It captures interactions and non-linear dependencies that are subtler and complex than
can be reflected by a linear model.
✔✔Model evaluation - ✔✔is primarily concerned with determining the predictive power
and relative error of a model.
✔✔Validation of the model - ✔✔primarily involves verifying that the model is robust, not
over fitted, and thus can be relied upon to maintain its predictive power for some time
into the future.
✔✔Model Implementation and Monitoring - ✔✔Integration and end-to-end testing
Reason codes for decisions
Monitoring
Model hold-outs
✔✔Misstatement of Age/Gender provision is - ✔✔a contractual provision found in most
life insurance policies.
✔✔Generally, if the insurer does not discover the age/gender misstatement until after
the death - ✔✔-the amount payable is adjusted to the amount the premiums paid would
have purchased at the correct age or sex according to the company premium rates at
the time of issue.
ANSWERS SET A+
✔✔In triage underwriting - ✔✔rather than obtaining an underwriting requirement
because it is indicated on an age-and-amount chart, the score from a predictive model
is used to determine which underwriting requirements are needed, including whether
certain requirements are needed at all.
✔✔Propensity Scoring - ✔✔where cases have more detailed info collected, such as lab
values, then more complex models can be built to estimate the likelihood or propensity
of an insured having a certain condition or disease.
✔✔Benefits of Predictive Modeling in Underwriting - ✔✔Improved mortality and more
competitive pricing
Faster case processing and less invasive underwriting
Lower underwriting costs
Better underwriter utilization
✔✔Challenges with Predictive Modeling for Underwriting - ✔✔Data availability
Data quality
Model fitting and subject matter expertise
✔✔A build dataset - ✔✔that is used to develop the model
✔✔A validation dataset - ✔✔that is used to test the model
✔✔Blind spots - ✔✔involve a model that is not predictive for certain age banks or other
segments of the population since the model had no basis in its build data set.
✔✔Linear Regression Model - ✔✔may be the most frequently used type of model to
represent business process.
, ✔✔Logistic Regression - ✔✔is similar to a linear regression model with the important
difference being that the target variable is binary (yes/no) in nature.
✔✔Cox proportional hazard model - ✔✔is the most widely used statistical technique for
estimating individual risk in studies of survival.
✔✔Decision Tree Model - ✔✔divides a dataset into progressively smaller sub-
segments, using the features in rules.
✔✔Decision Tree Model Advantage - ✔✔It has the advantage of being simple to
understand and interpret, however, it may not generalize well from the training data,
resulting in overfitting.
✔✔Random Forests - ✔✔Use multiple decision trees based on different subsets of the
data and/or different subsets of features, and outputs the mode or mean prediction of
the individual tree.
✔✔Random Forests Benefits - ✔✔This method helps to reduce the decision trees'
limitation of overfitting.
It captures interactions and non-linear dependencies that are subtler and complex than
can be reflected by a linear model.
✔✔Model evaluation - ✔✔is primarily concerned with determining the predictive power
and relative error of a model.
✔✔Validation of the model - ✔✔primarily involves verifying that the model is robust, not
over fitted, and thus can be relied upon to maintain its predictive power for some time
into the future.
✔✔Model Implementation and Monitoring - ✔✔Integration and end-to-end testing
Reason codes for decisions
Monitoring
Model hold-outs
✔✔Misstatement of Age/Gender provision is - ✔✔a contractual provision found in most
life insurance policies.
✔✔Generally, if the insurer does not discover the age/gender misstatement until after
the death - ✔✔-the amount payable is adjusted to the amount the premiums paid would
have purchased at the correct age or sex according to the company premium rates at
the time of issue.
