Decision List Model Options

Model name. You can generate the model name automatically based on the target or ID field (or model type in cases where no such field is specified) or specify a custom name.

Use partitioned data. If a partition field is defined, this option ensures that data from only the training partition is used to build the model. 

Create split models. Builds a separate model for each possible value of input fields that are specified as split fields. See Building Split Models for more information.

Mode. Specifies the method used to build the model.

• Generate model. Automatically generates a model on the models palette when the node is executed. The resulting model can be added to streams for purposes of scoring but cannot be further edited.
• Launch interactive session. Opens the interactive Decision List Viewer modeling (output) window, enabling you to pick from multiple alternatives and repeatedly apply the algorithm with different settings to progressively grow or modify the model. See the topic Decision List Viewer for more information.
• Use saved interactive session information. Launches an interactive session using previously saved settings. Interactive settings can be saved from the Decision List Viewer using the Generate menu (to create a model or modeling node) or the File menu (to update the node from which the session was launched).

Target value. Specifies the value of the target field that indicates the outcome you want to model. For example, if the target field churn is coded 0 = no and 1 = yes, specify 1 to identify rules that indicate which records are likely to churn.

Find segments with. Indicates whether the search for the target variable should look for a High probability or Low probability of occurrence. Finding and excluding them can be a useful way to improve your model and can be particularly useful when the remainder has a low probability.

Maximum number of segments. Specifies the maximum number of segments to return. The top N segments are created, where the best segment is the one with the highest probability or, if more than one model has the same probability, the highest coverage. The minimum allowed setting is 1; there is no maximum setting.

Minimum segment size. The two settings below dictate the minimum segment size. The larger of the two values takes precedence. For example, if the percentage value equates to a number higher than the absolute value, the percentage setting takes precedence.

• As percentage of previous segment (%). Specifies the minimum group size as a percentage of records. The minimum allowed setting is 0; the maximum allowed setting is 99.9.
• As absolute value (N). Specifies the minimum group size as an absolute number of records. The minimum allowed setting is 1; there is no maximum setting.

Segment rules.

Maximum number of attributes. Specifies the maximum number of conditions per segment rule. The minimum allowed setting is 1; there is no maximum setting.

• Allow attribute re-use. When enabled, each cycle can consider all attributes, even those that have been used in previous cycles. The conditions for a segment are built up in cycles, where each cycle adds a new condition. The number of cycles is defined using the Maximum number of attributes setting.

Confidence interval for new conditions (%). Specifies the confidence level for testing segment significance. This setting plays a significant role in the number of segments (if any) that are returned as well as the number-of-conditions-per-segment rule. The higher the value, the smaller the returned result set. The minimum allowed setting is 50; the maximum allowed setting is 99.9.