Logistic Regression

image

Logistic Regression Learner

Signals

Inputs:

Data set - Preprocessor

Preprocessed data

Outputs:

The logistic regression learning algorithm with settings as specified in the dialog.

Trained classifier (a subtype of Classifier). Logistic Regression Classifier sends data only if the data input is present.

Description

Logistic Regression Widget

  1. A name under which the learner appears in other widgets. The default name is 'Logistic regression'.
  2. The penalty type (either L1 or L2).
  3. Regularization value (higher value means less regularization).
  4. 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.

image