Module emd 

Empirical Mode Decomposition (EMD) for non-stationary signal analysis

EMD is a data-driven method for decomposing signals into Intrinsic Mode Functions (IMFs). Unlike wavelet transforms, EMD doesn’t require predefined basis functions and adapts to the local characteristics of the signal.

Financial Applications

• Trend extraction: Separate long-term trends from short-term fluctuations
• Volatility decomposition: Multi-scale volatility analysis
• Cycle detection: Identify market cycles without assuming periodicity
• Noise removal: Extract market microstructure noise
• Regime identification: Detect changes in market dynamics

EEMD Random Seed Configuration

The Ensemble EMD (EEMD) implementation supports configurable random seeds:

• Production use: Set random_seed: None for true randomness (default)
• Testing/Research: Use with_seed(seed) or set random_seed: Some(seed) for reproducibility

use ferro_wave::transform::EEMD;

// For production - uses random seed
let eemd = EEMD::default();

// For testing/research - reproducible results
let eemd_reproducible = EEMD::default().with_seed(42);

Structs

EEMD

Ensemble Empirical Mode Decomposition (EEMD) for improved robustness

EMDConfig

Configuration for EMD decomposition

EMDResult

EMD result containing all IMFs and the residue

IMF

Intrinsic Mode Function (IMF) - a single component from EMD

Enums

EndEffectStrategy

Strategy used to mitigate envelope end effects during EMD sifting.

Functions

emd

Perform Empirical Mode Decomposition

hilbert_transform

Compute the Hilbert transform of a real signal via FFT.