Tree vs Forest vs MARS vs GANs: Uncertainty, Intervals & Distributions

Hover or Drag Canvas: x = 0.00 | ŷ = 0.00 | ±0.00 (90% PI)

🎙️ OVERVIEW AUDIO GUIDE 1. Contender Architecture

00:00 / 02:40

Active Model Paradigm: RANDOM FOREST

Display Layer: PRED. INTERVAL (90%)

Synthetic Data Manifold: HETEROSCEDASTIC

Observation Count (N): 120 pts

Noise Scale (σ_ε): 1.00×

Model Complexity: Depth: 4

Coverage Target (1 - α): 90%

Epistemic (σ_ep)

0.082

Aleatoric (σ_al)

0.341

Empirical Coverage

89.4%

Empirical RMSE

0.298

Laboratory Gestures & Controls

Direct Canvas Gestures:

Left Click & Drag: Drag existing observations to alter local topology.
Shift + Click: Spawn custom outlier points at arbitrary coordinates.
Hover Cursor: Sweep across $x$ to render live dynamic violin distributions and drive carrier sonification.

Multi-Modal Sonification: In Bimodal Fork mode, the synthesizer triggers a dual harmonic dyad mapping both branches simultaneously!