Logistic Regression¶
Logistic Regression Node icon
The Clario Logistic Regression node uses logistic regression to build a model for a binary (2 class) dependent attribute. The model outcome is a prediction of the likelihood to belong to one of the two classes. The node connector can be connected to a variety of nodes, (e.g. Read File, Aggregate, Append, Missing, etc.), but requires a valid stream of data.
Configuration¶
The Logistic Regression node has three configuration tabs, Dependent Attribute, Weight Attribute, and Predictor Attributes.
Dependent Attribute Tab¶
Dependent Attribute tab
The Dependent Attribute tab contains an Available Attribute list box, a Dependent Attribute field, and a Settings area. First select an attribute from the Available Attributes list box and drag and drop it into the Dependent Attribute area (required). This attribute cannot be null. Under Settings, enter both Success and Failure values for the dependent attribute. Next, choose the Attribute Selection Method. Choices for the Attribute Selection Method are None and Stepwise. If Stepwise is chosen, you must also specify the ‘Maximum p to Enter’ and the ‘Minimum p to Remove’ values for the stepwise regression. See tips on Finding and Selecting Attributes.
Weight Attribute Tab¶
Weight Attribute tab
The Weight Attribute tab contains an Available Attributes list box, and a Weight Attribute box. This optional setting, whose default value is Null, provides a way to emphasize or de-emphasize certain observations in the regression. The attribute used for a Weight Attribute must be defined previously in another node, such as a Transform node, and consists of a numerical value representing the level of importance of each observation. To utilize the Weight Attribute, select the attribute from the Available Attributes list box and drag and drop it into the Weight Attribute box. Once an attribute is selected for the Weight Attribute, it is no longer available in the Available Attribute list boxes in the Dependent Attribute and Predictor Attribute tabs.
Predictor Attributes¶
The Predictor Attributes tab involves selecting the desired attribute(s) to use as predictor variables in the logistic regression. The method of selecting attributes will depend on which option was selected as the Selection Method on the Dependent Attribute configuration tab.
If Selection Method is ‘None’, attributes must be selected for entry into the model. Select the attribute(s) by clicking on it in the Available Attributes box and drag and drop into the Force Entry Attributes box. At least one attribute must be selected as a Force Entry Attribute.
Predictor Attributes tab
If Selection Method is ‘Stepwise’, select the attribute(s) by clicking on it in the Available Attributes box and drag and drop into the Candidate Attributes box. If there are any attributes that you wish to force into the model, select the attribute(s) from the Available Attributes list box and drag and drop it into the Force Entry Attributes box. At least one attribute must be selected as either a Force Entry or Candidate Attribute.
Results¶
There is one results set with two tabs (Response Profile and Detailed Results) for the Logistic Regression node. If Selection Method is ‘Stepwise’, an additional Step History tab is shown.
Response Profile Tab¶
This tab contains statistics such as Success Frequency, Failure Frequency, Missing number of rows, Success Weight, and Failure Weight.
Response Profile tab
Detailed Results Tab¶
This tab contains various statistics including Model Fit Statistics (AIC, SC, and neg2LogL), and Global Fit Statistics (Degrees of Freedom, Chi-square), and Analysis of Maximum Likelihood Estimates for each parameter (model estimates, model coefficient, degrees of freedom, standard error, Chi-square).
Detailed Results tab
Step History Tab (for stepwise method only)¶
This tab contains one row of data for each step in the model building process. Each step lists the attribute entered or removed at that step along with the resulting model R2.
Step History tab
Output Stream¶
The results from the Logistic Regression node can be read into Write, Score, and Evaluate. The results tables can also be exported into Excel by clicking the Export to Spreadsheet button found on the Toolbar. If the Logistic Regression results are written to a file to be used in a scoring application, make sure ‘Full Precision’ is selected as the number format to avoid truncation of model coefficients.
