Naive Bayes

A fast and simple probabilistic classifier based on Bayes’ theorem with the assumption of feature independence.

Signals

Inputs:

• Data

  A data set

• Preprocessor

  Preprocessing method(s)

Outputs:

• Learner

  A naive bayes learning algorithm with settings as specified in the dialog.

• Model

  A trained classifier. Output signal sent only if input Data is present.

Description

Naive Bayes learns a Naive Bayesian model from the data.

It only works for classification tasks.

This widget has two options: the name under which it will appear in other widgets and producing a report. The default name is Naive Bayes. When you change it, you need to press Apply.

Examples

Here, we present two uses of this widget. First, we compare the results of the Naive Bayes with another model, the Random Forest. We connect iris data from File to Test&Score. We also connect Naive Bayes and Random Forest to Test & Score and observe their prediction scores.

The second schema shows the quality of predictions made with Naive Bayes. We feed the Test&Score widget a Naive Bayes learner and then send the data to the Confusion Matrix. We also connect Scatterplot with File. Then we select the misclassified instances in the Confusion Matrix and show feed them to Scatterplot. The bold dots in the scatterplot are the misclassified instances from Naive Bayes.