PREDICTIVE ANALYTICS (PA) – ACTUARIAL
PRACTICE EXAM NEWEST 2025- 2026 UPDATE
ACTUAL 100 QUESTIONS & 100% CORRECT
ANSWERS GRADED A+ (BRAND NEW!!)
MOST TESTED TOPICS:
• Regression Modeling (Linear, Logistic, Poisson)
• Model Selection and Validation Techniques
• Time Series Analysis and Forecasting
• Loss Modeling and Risk Assessment
• Machine Learning Applications in Insurance
1. An insurance company wants to predict claim frequency using
historical policyholder data. They fit a Poisson regression model.
Which assumption is critical for the validity of this model?
A. The response variable is normally distributed
B. The variance is constant across all observations
C. Observations are dependent
D. The mean equals the variance of the response variable
Correct Answer: D
Poisson regression assumes that the mean and variance of the count
response are equal; violating this can lead to overdispersion.
, 2. A dataset shows that premium increases are correlated with
policyholder age. To avoid multicollinearity when building a
regression model, which action is most appropriate?
A. Include all correlated predictors
B. Remove or combine correlated predictors
C. Ignore correlation if p-values are significant
D. Use only categorical variables
Correct Answer: B
Multicollinearity can inflate standard errors and destabilize coefficient
estimates; removing or combining correlated variables mitigates this.
3. A predictive model for claims predicts 100 claims, while the actual
observed claims are 120. Which metric measures this prediction
accuracy?
A. ROC AUC
B. Log-likelihood
C. Mean Absolute Error (MAE)
D. Chi-square statistic
Correct Answer: C
MAE quantifies the average absolute difference between predicted and
actual values, making it appropriate for continuous count predictions.
4. An actuary is modeling claim severity with highly skewed data.
Which transformation is most appropriate to stabilize variance?
A. Square
B. Logarithm
, C. Reciprocal
D. Cube
Correct Answer: B
Log transformation reduces skewness and stabilizes variance for right-
skewed claim severity distributions.
5. A dataset has 10,000 policies, but only 1% report a claim. Which
technique helps address this class imbalance in a predictive
model?
A. Ignore imbalance
B. Feature scaling
C. Oversampling the minority class
D. Principal component analysis
Correct Answer: C
Oversampling or SMOTE helps models learn patterns from rare events
without biasing toward the majority class.
6. When comparing two Poisson regression models, which statistic
evaluates model fit while penalizing for model complexity?
A. R-squared
B. p-value
C. Akaike Information Criterion (AIC)
D. Standard deviation
Correct Answer: C
AIC balances goodness-of-fit with model complexity, helping choose the
more parsimonious model.
, 7. A car insurer wants to forecast monthly claims using historical
data with trend and seasonality. Which model is most suitable?
A. Linear regression
B. Logistic regression
C. ARIMA with seasonal components
D. Decision tree
Correct Answer: C
ARIMA with seasonal terms (SARIMA) captures both trend and
seasonality in time series data.
8. Which statement about logistic regression is correct?
A. Dependent variable must be continuous
B. Coefficients are estimated using least squares
C. Coefficients represent the log odds of the outcome
D. It cannot include categorical predictors
Correct Answer: C
Logistic regression models log odds; coefficients quantify the change in
log odds per unit change in predictors.
9. An actuary fits a generalized linear model for claim severity. The
residual deviance is substantially higher than the degrees of
freedom. What does this indicate?
A. Good model fit
B. Overdispersion
PRACTICE EXAM NEWEST 2025- 2026 UPDATE
ACTUAL 100 QUESTIONS & 100% CORRECT
ANSWERS GRADED A+ (BRAND NEW!!)
MOST TESTED TOPICS:
• Regression Modeling (Linear, Logistic, Poisson)
• Model Selection and Validation Techniques
• Time Series Analysis and Forecasting
• Loss Modeling and Risk Assessment
• Machine Learning Applications in Insurance
1. An insurance company wants to predict claim frequency using
historical policyholder data. They fit a Poisson regression model.
Which assumption is critical for the validity of this model?
A. The response variable is normally distributed
B. The variance is constant across all observations
C. Observations are dependent
D. The mean equals the variance of the response variable
Correct Answer: D
Poisson regression assumes that the mean and variance of the count
response are equal; violating this can lead to overdispersion.
, 2. A dataset shows that premium increases are correlated with
policyholder age. To avoid multicollinearity when building a
regression model, which action is most appropriate?
A. Include all correlated predictors
B. Remove or combine correlated predictors
C. Ignore correlation if p-values are significant
D. Use only categorical variables
Correct Answer: B
Multicollinearity can inflate standard errors and destabilize coefficient
estimates; removing or combining correlated variables mitigates this.
3. A predictive model for claims predicts 100 claims, while the actual
observed claims are 120. Which metric measures this prediction
accuracy?
A. ROC AUC
B. Log-likelihood
C. Mean Absolute Error (MAE)
D. Chi-square statistic
Correct Answer: C
MAE quantifies the average absolute difference between predicted and
actual values, making it appropriate for continuous count predictions.
4. An actuary is modeling claim severity with highly skewed data.
Which transformation is most appropriate to stabilize variance?
A. Square
B. Logarithm
, C. Reciprocal
D. Cube
Correct Answer: B
Log transformation reduces skewness and stabilizes variance for right-
skewed claim severity distributions.
5. A dataset has 10,000 policies, but only 1% report a claim. Which
technique helps address this class imbalance in a predictive
model?
A. Ignore imbalance
B. Feature scaling
C. Oversampling the minority class
D. Principal component analysis
Correct Answer: C
Oversampling or SMOTE helps models learn patterns from rare events
without biasing toward the majority class.
6. When comparing two Poisson regression models, which statistic
evaluates model fit while penalizing for model complexity?
A. R-squared
B. p-value
C. Akaike Information Criterion (AIC)
D. Standard deviation
Correct Answer: C
AIC balances goodness-of-fit with model complexity, helping choose the
more parsimonious model.
, 7. A car insurer wants to forecast monthly claims using historical
data with trend and seasonality. Which model is most suitable?
A. Linear regression
B. Logistic regression
C. ARIMA with seasonal components
D. Decision tree
Correct Answer: C
ARIMA with seasonal terms (SARIMA) captures both trend and
seasonality in time series data.
8. Which statement about logistic regression is correct?
A. Dependent variable must be continuous
B. Coefficients are estimated using least squares
C. Coefficients represent the log odds of the outcome
D. It cannot include categorical predictors
Correct Answer: C
Logistic regression models log odds; coefficients quantify the change in
log odds per unit change in predictors.
9. An actuary fits a generalized linear model for claim severity. The
residual deviance is substantially higher than the degrees of
freedom. What does this indicate?
A. Good model fit
B. Overdispersion
