746 お客様のコメント
質問 1：
The number of hidden layers in this Neural Network model is which of the following?
Response:
A. 1
B. 3
C. 4 or more
D. 2
正解：A
質問 2：
Perform these tasks in SAS Enterprise Miner:
* Continue to use the same diagram. Define and create the data set CREDIT_SCORE for scoring. The variables (their roles and measurement levels) in the CREDIT_SCORE data should be set as identical to those in the CREDIT dat a. The only exception is that the scoring data does not have a TARGET variable.
* Find the best model out of Decision Tree, Decision Tree (3-way), Regression, and Neural Network as defined by each of the four model's overall performance in the validation data measured by average squared error. Now, use this best model to score the CREDIT_SCORE data.
CREDIT SCORE:
The distribution of the predicted probabilities of TARGET=0 in the scoring data is approximately which of the following?
Response:
A. bimodal
B. normal
C. right skewed
D. left skewed
正解：D
質問 3：
You are building a model to identify fraud. Your model will produce predictions that can be interpreted as the probability of fraud. You will pass on the top 100 scoring cases to management for investigation. Assume that you have sufficient data to hold out a validation and test data set for model evaluation.
Which selection would represent a reasonable ordering of fit statistics (best to worst) for this situation?
Select one:
Response:
A. R-square, AIC, K-S statistic
B. Misclassification rate, ROC index, ASE
C. ASE, Lift for the top 10%, ROC index
D. ROC index, ASE, misclassification rate
正解：D
質問 4：
An analyst is performing a market basket analysis (affinity analysis) on the purchase of Shaving Cream and Seltzer Water. The purchase data from a set of 250 customers is shown below:
What is the confidence of the rule "Shaving Cream implies Seltzer Water"? You may use a calculator for this question. On the certification exam, an on-screen calculator is provided for you.
Select one:
Response:
A. 60%
B. 40%
C. 57%
D. 67%
正解：C
質問 5：
-> Add a Decision Tree node after the Impute node with TARGET as the dependent variable and all other input variables as independent variables (main effects only).
- Allow for 1 substitute rule in case the variable for the primary splitting rule is missing.
- Disable pruning for the decision tree.
-> Add another Neural Network node after the decision tree with TARGET as the dependent variable and all other input variables as independent variables (main effects only).
- Configure the Neural Network model to use Average Error for Model Selection Criterion.
-> Run the process flow.
What is the number of input variables being used by the Neural Network Model?
Enter your numeric answer in the space below:
Response:
A. 16
B. 13
C. 11
D. 10
正解：A
質問 6：
What is the kurtosis value for the variable TLDel60Cnt24?
Response:
A. between 14 and 16.99
B. between 10 and 13.99
C. 17 or higher
D. less than 10
正解：A
質問 7：
Look over the output from the Neural Network model. Which of the following statement(s) is (are) true?
Response:
A. All of the above
B. The optimization for the model has not been completed.
C. The misclassification error for the test data is 0.154255.
D. The model has too few input variables.
正解：B
The number of hidden layers in this Neural Network model is which of the following?
Response:
A. 1
B. 3
C. 4 or more
D. 2
正解：A
質問 2：
Perform these tasks in SAS Enterprise Miner:
* Continue to use the same diagram. Define and create the data set CREDIT_SCORE for scoring. The variables (their roles and measurement levels) in the CREDIT_SCORE data should be set as identical to those in the CREDIT dat a. The only exception is that the scoring data does not have a TARGET variable.
* Find the best model out of Decision Tree, Decision Tree (3-way), Regression, and Neural Network as defined by each of the four model's overall performance in the validation data measured by average squared error. Now, use this best model to score the CREDIT_SCORE data.
CREDIT SCORE:
The distribution of the predicted probabilities of TARGET=0 in the scoring data is approximately which of the following?
Response:
A. bimodal
B. normal
C. right skewed
D. left skewed
正解：D
質問 3：
You are building a model to identify fraud. Your model will produce predictions that can be interpreted as the probability of fraud. You will pass on the top 100 scoring cases to management for investigation. Assume that you have sufficient data to hold out a validation and test data set for model evaluation.
Which selection would represent a reasonable ordering of fit statistics (best to worst) for this situation?
Select one:
Response:
A. R-square, AIC, K-S statistic
B. Misclassification rate, ROC index, ASE
C. ASE, Lift for the top 10%, ROC index
D. ROC index, ASE, misclassification rate
正解：D
質問 4：
An analyst is performing a market basket analysis (affinity analysis) on the purchase of Shaving Cream and Seltzer Water. The purchase data from a set of 250 customers is shown below:
What is the confidence of the rule "Shaving Cream implies Seltzer Water"? You may use a calculator for this question. On the certification exam, an on-screen calculator is provided for you.
Select one:
Response:
A. 60%
B. 40%
C. 57%
D. 67%
正解：C
質問 5：
-> Add a Decision Tree node after the Impute node with TARGET as the dependent variable and all other input variables as independent variables (main effects only).
- Allow for 1 substitute rule in case the variable for the primary splitting rule is missing.
- Disable pruning for the decision tree.
-> Add another Neural Network node after the decision tree with TARGET as the dependent variable and all other input variables as independent variables (main effects only).
- Configure the Neural Network model to use Average Error for Model Selection Criterion.
-> Run the process flow.
What is the number of input variables being used by the Neural Network Model?
Enter your numeric answer in the space below:
Response:
A. 16
B. 13
C. 11
D. 10
正解：A
質問 6：
What is the kurtosis value for the variable TLDel60Cnt24?
Response:
A. between 14 and 16.99
B. between 10 and 13.99
C. 17 or higher
D. less than 10
正解：A
質問 7：
Look over the output from the Neural Network model. Which of the following statement(s) is (are) true?
Response:
A. All of the above
B. The optimization for the model has not been completed.
C. The misclassification error for the test data is 0.154255.
D. The model has too few input variables.
正解：B
西尾** -
飽きない工夫もされているので、A00-255独学等の良い相棒になってくれると思います。