SOA Exam PA Review Questions &
actual answers.
Explain the difference between descriptive, predictive, and prescriptive modeling ANS -Descriptive -
focuses on what happened in the past and aims to describe the observed trends by identifying
relationships between variables.
Predictive - What will happen in the future and making accurate predictions.
Prescriptive - Combo of optimization and simulation to investigate the impact of prescribed actions in
different scenarios.
Explain the difference between supervised and unsupervised learning ANS -Supervised learning has a
target variable and the goal is to understand the relationship between it and the predictors or make
predictions. Unsupervised learning has no target variable, and we are interested in relationships and
structures between variables.
Explain how stratified sampling contributes to a more representative sample than random sampling ANS
-Divides the population into non-overlapping groups in a non-random fashion. Allows for all stratum to
be represented in collected data.
Explain three data quality issues one should examine in practice ANS -Reasonableness - data values need
to be reasonable in their recording.
Consistency - data must be inputted consistently with the same levels, units, categorical variables
recorded in consistent manner.
Sufficient documentation - Have overall description, variable description, updates, accountability, and
how to manage the dataset.
Explain the problem with target leakage in predictive analytics ANS -Predictors that leak information
about the target variable but will not be available when the model is applied in practice.
Explain how to use a time variable to make the training/test set split and the advantage of doing so ANS
-Allocate the older observations to the training set and the more recent to the test set. This is useful for
evaluating how well a model extrapolates time trends observed in the part to future, unseen years.
, Explain what hyperparameters are and why they are important for a predictive model ANS -
Hyperparameters are values that control some aspect of the fitting process. Typically involves optimizing
an objective function or in the set of constraints. Supply their values in advance.
Explain the difference between bias and variance in a predictive analytic context ANS -Bias is the
difference between the expected value of our fitted function and the true value of the signal function.
Variance quantifies the amount by which our fitted function would change if we estimated it using a
different training set.
Explain the difference between variables and features in a predictive analytic context ANS -Variables are
a predictor in a model or a raw measurement before any transformations are applied. Features are
derivations from the original variables and provide an alternative view of the information.
Explain the difference between dimensionality and granularity ANS -Dimensionality is the number of
possible levels that a variable has and is specific to categorical variables. Granularity is how fine the
information is stored (numeric and categorical). Not always possible to order by granularity.
Explain the problem with RSS and R2 as model selection measures ANS -RSS and R2 favor more complex
linear models which fit the training data better but are not a fair reflection of model quality. Lowest
training error but not necessarily the lower test error.
Explain the rationale behind and the difference between the AIC and BIC ANS -They are indirect
measures of prediction performance because they adjust a goodness-of-fit measure like RSS to account
for model complexity. The penalty term for BIC is higher for almost all reasonable value of training set
size.
Explain the advantages and disadvantages of polynomial regression ANS -An advantage of polynomial
regression is its ability to model non-linear and more complex relationships. A disadvantage is that
polynomial regression is difficult to interpret in some contexts. Choice of m is often an iterative exercise
because there is no simple rule to determine its value.
Explain the meaning of interaction ANS -An interaction between two variables means that the
association between one predictor and the target variable depends on the value or level of another
predictor. In predictive analytics, an interaction is interpreted as a change in X1 with respect to Y
changes in value with respect to X2.
actual answers.
Explain the difference between descriptive, predictive, and prescriptive modeling ANS -Descriptive -
focuses on what happened in the past and aims to describe the observed trends by identifying
relationships between variables.
Predictive - What will happen in the future and making accurate predictions.
Prescriptive - Combo of optimization and simulation to investigate the impact of prescribed actions in
different scenarios.
Explain the difference between supervised and unsupervised learning ANS -Supervised learning has a
target variable and the goal is to understand the relationship between it and the predictors or make
predictions. Unsupervised learning has no target variable, and we are interested in relationships and
structures between variables.
Explain how stratified sampling contributes to a more representative sample than random sampling ANS
-Divides the population into non-overlapping groups in a non-random fashion. Allows for all stratum to
be represented in collected data.
Explain three data quality issues one should examine in practice ANS -Reasonableness - data values need
to be reasonable in their recording.
Consistency - data must be inputted consistently with the same levels, units, categorical variables
recorded in consistent manner.
Sufficient documentation - Have overall description, variable description, updates, accountability, and
how to manage the dataset.
Explain the problem with target leakage in predictive analytics ANS -Predictors that leak information
about the target variable but will not be available when the model is applied in practice.
Explain how to use a time variable to make the training/test set split and the advantage of doing so ANS
-Allocate the older observations to the training set and the more recent to the test set. This is useful for
evaluating how well a model extrapolates time trends observed in the part to future, unseen years.
, Explain what hyperparameters are and why they are important for a predictive model ANS -
Hyperparameters are values that control some aspect of the fitting process. Typically involves optimizing
an objective function or in the set of constraints. Supply their values in advance.
Explain the difference between bias and variance in a predictive analytic context ANS -Bias is the
difference between the expected value of our fitted function and the true value of the signal function.
Variance quantifies the amount by which our fitted function would change if we estimated it using a
different training set.
Explain the difference between variables and features in a predictive analytic context ANS -Variables are
a predictor in a model or a raw measurement before any transformations are applied. Features are
derivations from the original variables and provide an alternative view of the information.
Explain the difference between dimensionality and granularity ANS -Dimensionality is the number of
possible levels that a variable has and is specific to categorical variables. Granularity is how fine the
information is stored (numeric and categorical). Not always possible to order by granularity.
Explain the problem with RSS and R2 as model selection measures ANS -RSS and R2 favor more complex
linear models which fit the training data better but are not a fair reflection of model quality. Lowest
training error but not necessarily the lower test error.
Explain the rationale behind and the difference between the AIC and BIC ANS -They are indirect
measures of prediction performance because they adjust a goodness-of-fit measure like RSS to account
for model complexity. The penalty term for BIC is higher for almost all reasonable value of training set
size.
Explain the advantages and disadvantages of polynomial regression ANS -An advantage of polynomial
regression is its ability to model non-linear and more complex relationships. A disadvantage is that
polynomial regression is difficult to interpret in some contexts. Choice of m is often an iterative exercise
because there is no simple rule to determine its value.
Explain the meaning of interaction ANS -An interaction between two variables means that the
association between one predictor and the target variable depends on the value or level of another
predictor. In predictive analytics, an interaction is interpreted as a change in X1 with respect to Y
changes in value with respect to X2.
