Module time_series_analysis 

Time-series analysis utilities: statistics, filtering, autocorrelation, trend decomposition, spectral analysis, change-point detection, and AR models.

Structs

ArModel

Autoregressive model of order p.

Autocorrelation

Autocorrelation function (ACF) estimator.

ChangePointDetection

CUSUM-based change-point detector.

CrossCorrelation

Cross-correlation estimator for two equal-length signals.

Periodogram

Lomb-Scargle periodogram for unevenly-sampled data.

TimeSeries

A time series with optional explicit timestamps.

TimeSeriesStats

Descriptive statistics computed over a TimeSeries.

Enums

TimeSeriesFilter

Filter types available through apply_filter.

Functions

adf_test

Simplified Augmented Dickey-Fuller test for stationarity.

apply_filter

Apply a TimeSeriesFilter to data and return the filtered signal.

compute_power_spectrum

Compute the one-sided power spectral density via a direct DFT.

compute_stats

Compute descriptive statistics for a TimeSeries.

decompose_trend

Classical additive decomposition of a time series into trend, seasonal, and residual components.

ema

Compute an exponential moving average with smoothing factor alpha ∈ (0, 1].

find_peaks

Find local maxima whose prominence exceeds min_prominence.

hurst_exponent

Estimate the Hurst exponent of a time series using R/S (rescaled range) analysis.

interpolate_linear

Fill NaN values in data using linear interpolation between the nearest valid neighbours. Leading and trailing NaNs are filled with the first / last valid value respectively.

interpolate_spline

Fill NaN values in data using simple cubic-spline-like interpolation (Catmull-Rom tangents) between the nearest valid neighbours.

moving_average

Compute a centred (symmetric) simple moving average.

zero_crossing_rate

Compute the zero-crossing rate: fraction of consecutive-sample sign changes.