Sample Entropy Analysis Example

This section provides a practical guide to conducting a valid entropy analysis using the StationarySignals and SampleEntropy classes in this package. Below are the steps to accomplish this:

Preliminary Set-Up

import numpy as np
import pandas as pd
from eristropy.stationarity import StationarySiganls
from eristropy.sample_entropy import SampleEntropy

Suppose we have time series signals, here we create synthetic signals for demonstration purposes.

signal_ids = np.repeat(["signal_1", "signal_2"], 100)
timestamps = np.tile(np.arange(100), 2)
rng = np.random.default_rng(17)
signal_1_values = rng.uniform(-5, 5, size=(100,))
signal_2_values = rng.uniform(-5, 5, size=(100,))
values = np.concatenate((signal_1_values, signal_2_values))

df = pd.DataFrame({
    "signal_id": signal_ids,
    "timestamp": timestamps,
    "value": values
})

Ensuring Stationary Signals

Before computing SampEn, it's crucial to ensure that the signals are stationary (see Stationarity for more details). This is where StationarySignals comes in handy.

signals = StationarySignals(df, method='difference')
stationary_df = signals.make_stationary_signals()

In this example, we use the "differencing"-based approach, but we also allow users to de-trend the signals if desired.

Find Optimal SampEn Parameters & Compute SampEn

Now, that we have weakly stationary signals, we can compute the SampEn for all unique signals in the dataset. You can also estimate the uncertainty if needed.

sampen = SampleEntropy(stationary_df)
result_df = sampen.compute_all_sampen(optimize=True, estimate_uncertainty=True)

The result_df will contain SampEn estimates from the optimal \((m, r)\) combination for each unique signal, which you can then analyze further.

print(result_df)