Initializing help system before first use

Identifying Variables in Decision Trees

Xpress Insight identifies variables in a unique way specific to decision trees.

Variables that can be profiled through the tree are available in the Profiles panel. These variables can be categorized as either continuous or categorical. Categorical variables define the array of possible categories. Continuous variables are normally numeric values—note that a range of numeric values may also be classed as categorical. All categorical profile variables are treated as possible targets for the tree and are available for selection in the Target dialog.

The following table lists each data type available in decision trees and how it is used in Xpress Insight.
Data Type Identification Criteria Inserting Splits Target-Driven Decision Trees Statistics for Profiling Variables

real

All numeric variables

Enter branch thresholds.

Best Split algorithm is supported.

  • Categorical—Numeric variables with 10 or fewer unique values
  • Continuous—Numeric variables with more than 10 unique values

enum (enumeration)

All string variables with 100 or fewer unique values

All unique values are automatically added to the tree after you click APPLY. Each value is shown as its own node in the tree.

Best Split algorithm is not supported.

Categorical

string

All string variables with more than 100 unique values

Each unique value must be entered manually as a branch value.

Best Split algorithm is not supported.

Not available as a profile variable.
Variables in Decision Trees
Tip: You can verify the data type of a variable in a decision tree by right-clicking its level and selecting Properties to open its Properties dialog box.
Note: Any split on a string variable in a decision tree will contain an extra 'Any other value' branch to which the user will have to assign a treatment. This will be the case even for string variables with a finite set of known values (yes/no, hot/cold). This is stored as a FJDT special condition. Decision Trees that use string variables are not fully supported. For example, you cannot create them with TAO scenarios. It is also not possible to use Decision Trees that use the FJDT special condition “Any other value” as base trees in tree templates.
Covered Topics
  • Identifying Categorical and Continuous Variables
    In general, variables and data represent groups of things about categorical or discrete characteristics, or they represent measurements on a continuous scale. Examples of categorical variables are a person's occupation or the number of inquiries made for borrowers over the last 5 months. Examples of continuous variables are a person's income or the daily temperature of the ocean.
Parent Topic Manipulating Decision Trees

© 2001-2024 Fair Isaac Corporation. All rights reserved. This documentation is the property of Fair Isaac Corporation (“FICO”). Receipt or possession of this documentation does not convey rights to disclose, reproduce, make derivative works, use, or allow others to use it except solely for internal evaluation purposes to determine whether to purchase a license to the software described in this documentation, or as otherwise set forth in a written software license agreement between you and FICO (or a FICO affiliate). Use of this documentation and the software described in it must conform strictly to the foregoing permitted uses, and no other use is permitted.