Naive Bayes classifies using Bayes' probability theorem — under the simplifying assumption that all features are independent of each other. This app makes the conditional probabilities directly visible in feature space.
The app shows how the Naive Bayes classifier estimates conditional probabilities from training data and derives a decision boundary from them. For simplicity, the two-dimensional feature space is divided into four quadrants: left/right (feature x) and top/bottom (feature y).
The conditional frequencies are shown at the edges of the diagram — how many points of each class lie in which half-space. Clicking on a quadrant reveals the corresponding Bayes formula and calculates which class is more probable for a point in that region.
Move the mouse over the labelled fields at the edge of the visualisation area to highlight the corresponding half of the feature space. Click on a quadrant to display the Bayes formula for that region.
Data points can be added or removed — the conditional probabilities and the decision boundary adjust immediately.
Interested in AI visualisations for teaching?
The VisualApps are created as a teaching and transfer project at Reutlingen University and are used in corporate training and talks.
Get in touch