๐ŸŽพ

Zepp/Watch

Bulkhead ฯ„ Physics Microscope
← Back to Lab
๐Ÿ“Š Evidence Reproducible / Statistical Audit

Forensic Data Flow

Zepp DB
→
Debug Features
→
Classifier / Correlation
→
Phoenix Model
→
Apple Watch Comparison

Extracting physical meaning from legacy sensor features to reconcile 2014 swing semantics with 2026 biomechanical reconstructions.

Stroke Classification

97%
Prediction Accuracy

The model identifies stroke types with 97% accuracy using only legacy debug signals.

Top Feature: dbg_sum_gx (31.5% weight)

Apple Watch Integration

Status: Pairing Scaffold

Identified broad temporal overlap for session 2026-01-30 (97 Zepp impacts within watch window).

Note: Currently establishes a pairing scaffold with broad temporal overlap. Strict physical impact alignment is pending.

Tier 3: The Semantic Layer

The Oracle's Verdict

Interactive Correlation Matrix (r=0.84 for dbg_var_2) Open Fullscreen →
Supplemental Evidence

Stroke Separation Power

Visualizing dbg_sum_gx as the primary stroke differentiator. Open Fullscreen →
Tier 2: The Spatial Layer

The Physics Microscope

Interactive 3D Trajectory (Swing ID: 1658297433600) Open Fullscreen →
Tier 1: The Signal Layer

The Raw Truth

333Hz Multi-Axis Raw Sensor Trace Open Fullscreen →

"Legacy opaque features are physically interpretable."