| Use previous input description
|
Select this check box to use the variable selection specified in the Select dependent variables and predictors node, accessed via the ribbon bar or by double-clicking the data source node in the workspace. Note that the previous Select Cases and Weight specifications will be used also. If case selection conditions and/or weights were previously specified, you can view those specifications from this workspace dialog box, but you cannot change them (the options will be dimmed).
|
| Variables
|
Click this button to display a four-variable lists selection dialog box from which you can select the dependent variable, categorical predictor variables, continuous predictors, and count variable (optional).
From the first list, select the dependent variable. If you are analyzing a classification problem with a categorical response variable, the dependent variable must contain text or numeric (integer) codes that identify the class or group to which each case or observation belongs.
From the second list, select the categorical variables (factors). Note that these variables must contain text or numeric codes that identify the classes or groups to which each case or observation belongs.
From the third list, select the continuous predictors.
From the fourth list, select the count variable. This variable is optional. If selected, it will be used as a case-multiplier in the analysis. Thus, by using a count variable, you can analyze input data that has previously been aggregated (tabulated).
Use dependent as ordinal variable. Ordered Twoing Criterion is used. Select this check box to analyze a dependent (criterion) variable that is ordinal in nature. A brief description of how the ordered twoing criterion is used to split a node is given below.
At each node the following computations are performed.
- Denote the classes of the ordinal response variable as C = {1,2,...J}.
- Create two composite classes C1 and C2 = C-C1 such that C1 is of the form {1,2,...jn}.
- Compute the decrease in the impurity measure as though this were a two class problem using the Gini criterion (for the two class problem this is equivalent to the twoing criterion) for all possible splits.
- Find the classes C1 and C2 such that this decrease in the impurity measure is maximized.
|
