Return the differenced version of the series. Parameters ---------- series : pd.Series Original time series d : int

Detect the frequency of the time series. Parameters ---------- series : pd.Series Time series with datetime index

Detect seasonal period from ACF on a stationarized series. Key behavior: - Uses get_stationary_series(series, d) with proper differencin

Run ADF and KPSS tests for stationarity with verified differencing. Uses both tests for robustness: - ADF: null = non-stationary (re

Load and prepare time series data from CSV. Parameters ---------- filepath : str Path to CSV file date_col : str, op

Fallback variance check when Box-Cox cannot be applied. Uses mean-variance correlation across chunks.

Analyze ACF and PACF on stationary series.

Analyze data quality metrics.

Analyze distribution statistics.

Analyze seasonality using STL decomposition.

Analyze trend using STL decomposition with corrected threshold.

Check if a variance-stabilizing transform is needed using Box-Cox analysis. Analyzes optimal Box-Cox lambda: - λ ≈ 0: recommend log

Generate all diagnostic visualizations.

Generate human-readable summary file.

Load diagnostic state from diagnose.py if available.

Create ACF and PACF plots on stationary series.

Create box plots by time intervals based on data frequency.

Create STL decomposition plot.

Create histogram with distribution statistics.

Create lag scatter plots for visual autocorrelation inspection.

Create rolling statistics plot to visualize stationarity.

Create time series line plot.

Run complete diagnostic pipeline.

Test if series is white noise using Ljung-Box at multiple lags. Tests at lags 10, 20, and n/5 for robustness.