DCTTransform

transforms.DCTTransform(
    low_pass_size=5,
    output_length=100,
    scale_range=False,
    scale_values=False,
)

Compute a low-pass discrete cosine transform.

Parameters

Name	Type	Description	Default
low_pass_size	int	Number of DCT frequency components to retain. Lower values produce smoother curves. Defaults to 5. - 2-5: Very smooth, broad narrative patterns only - 5-10: Balanced smoothing (recommended for most uses) - 10-20: Preserves more detail, shows secondary peaks - >20: Minimal smoothing, may retain noise	5
output_length	int	Number of points in the interpolated output. Defaults to 100. Common values: 100-200 for visualization, 1000+ for analysis.	100
scale_range	bool	When True, rescale output to [-1, 1]. Mutually exclusive with scale_values. Use for comparing texts on a common scale.	False
scale_values	bool	When True, standardize output (mean=0, std=1). Mutually exclusive with scale_range. Use for statistical analysis.	False

Examples

>>> # Smooth smoothing for broad patterns
>>> dct_smooth = DCTTransform(low_pass_size=5, output_length=100)
>>> smoothed = dct_smooth.transform(raw_scores)
>>>
>>> # More detail, normalized range
>>> dct_detailed = DCTTransform(
...     low_pass_size=15,
...     output_length=200,
...     scale_range=True
... )
>>>
>>> # For statistical comparison across texts
>>> dct_stats = DCTTransform(scale_values=True)

Notes

You cannot set both scale_range and scale_values to True. Choose one based on your use case:

• Use scale_range=True for visualization and direct comparison
• Use scale_values=True for statistical analysis
• Use neither for preserving original scale