Logistic Regression
Logistic Regression Learner
Signals
Inputs:
- Data
Data set - Preprocessor
Preprocessed data
Outputs:
- Learner
The logistic regression learning algorithm with settings as specified in the dialog.
- Logistic Regression Classifier
Trained classifier (a subtype of Classifier). Logistic Regression Classifier sends data only if the data input is present.
Description
- A name under which the learner appears in other widgets. The default name is 'Logistic regression'.
- The penalty type (either L1 or L2).
- Regularization value (higher value means less regularization).
- Set numerical tolerance (permitted deviation from expected value).
Example
The widget is used just as any other widget for inducing a classifier. This is an example demonstrating the prediction value of logistic regress used on voting.tab data set. We first use a Logistic Regression learner to provide a LR classifier for Predictions widget. We want to see the quality of LR prediction model a person being a republic or a democrat based on their voting patterns. In Select Attributes we choose logistic regression as a feature and party as a class. Then we use Scatterplot to see which instances were correctly predicted and which were false.