ax-js Visualization Demos

Client-side Gaussian Process predictions mirroring Ax / BoTorch

Posterior Visualization

1D Slice Plots

See how each input affects your outcome. Interact to visually explore the space in one-dimensional slices.

2D Response Surface

See how two inputs jointly affect your outcome. Interact to visually explore the space in two-dimensional slices.

Model Diagnostics

LOO Cross-Validation

How well does the model predict the data? See what the model would predict if we didn't observe the data.

Scatter Plots

An overview of generic scatterplots available in ax-js.

Feature Importance

Which parameters matter most? Separates direct effects from interactions with other parameters.

Optimization Trace

Is the optimizer making progress? Shows each trial's outcome with a running best-so-far line.

Multi-Objective

Ax Explorer

Understand everything there is to know about how your inputs affect tradeoffs in your system.

Multi-Objective Radar

Radar chart for constrained multi-objective problems — objectives, constraints, feasibility.

Jupyter / Python

Jupyter Notebook

All diagnostic visualizations rendered inline — slice plots, response surfaces, cross-validation, feature importance. Pre-built outputs, no execution required.

IPython + ax-platform
Bayesian Optimization Loop

Live Bayesian Optimization

Watch optimization happen in real time — the model updates, picks the next point, and the surface evolves.

Preferential BO (PBO)

Use Bayesian optimization to learn your preferences.