Understanding the Decision Tree Panel The panel Decision Tree is split into three parts:
Note: There are 3 correct answers to this question.

• A. The decision tree itself, which is displayed in the upper section of the panel
• B. Two tabs located in the left bottom part of the panel provide you with information on the nodes and with the profit curve corresponding to the current decision tree
• C. A navigator allowing you to visualize what part of the tree you are studying is displayed in the right bottom part of the panel.\
• D. A consequence of 1 is that this metric is resistant to outliers: when there are a few occurrences of the target with very high values with respect to the rest of the target value distributions, the notion of normal profit is not impacted.

Once the model has been generated, you must verify its validity by examining the performance indicators:
Note: There are 2 correct answers to this question.

• A. The environment variable definition is using the POSIX standard shell sh semantic, which first defines the variable, then exports it. To guarantee a proper functioning, The environment variable definition is using the POSIX standard shell sh semantic, which first defines the variable, then exports it. To guarantee a proper functioning,
• B. The prediction confidence defines the degree of robustness of the model, that is, its capacity to achieve the same explanatory power when applied to a new dataset. In other words, the degree of robustness corresponds to the predictive power of the model when applied to an application dataset. A model with a prediction confidence inferior to 0.95 must be considered with caution. The performance of such a model is very likely to vary between the training dataset and the application datasets.
• C. The predictive power allows you to evaluate the explanatory power of the model, that is, its capacity to explain the target variable when applied to the training dataset. A perfect model possesses a predictive power of 1 and a completely random model possesses a predictive power of 0. No minimum threshold is required for the predictive power of a model. This depends upon the context of your work, that is, your domain of application, the nature of your data and your business issue. 26 P U B L I C Automated Analytics User Guides and Scenarios Modeling Concepts In some cases, a model with a predictive power as low as 0.1 may allow realization of a profit of several thousands dollars. In all cases, a positive predictive power indicates that the model generated will perform better than a random model.

You work for an automobile manufacturer and wish to send a
promotional mailing to your prospects. Modeler
Regression/Classification allows you to:
Note: There are 2 correct answers to this question.

• A. the user who runs SAP Predictive Analytics is also the one used to
  install the software
• B. Predict the response rate to such a mailing sent to new prospects.
• C. Understand why previous prospects responded to such a mailing,

What type of join is generated when creating
Analytical Data Set and Data Manipulation from the
Data Manager?
Note: There are 1 correct answers to this question.

• A. Cross join (Cartesian product)
• B. Right outer join
• C. Inner join
• D. Left outer join