Module point_process

Temporal point process features extracted from event timestamps.

The when of events is a signal independent of what the vectors contain. This module extracts features from inter-event intervals that characterize temporal patterns: regularity, burstiness, self-excitation, circadian rhythms.

Features

Feature	Range	Interpretation
burstiness	[−1, 1]	−1 = perfectly regular, 0 = Poisson, +1 = maximally bursty
memory	[−1, 1]	Autocorrelation of consecutive gaps
temporal_entropy	[0, ∞)	Entropy of gap distribution (higher = more irregular)
intensity_trend	ℝ	Slope of event rate over time (positive = accelerating)
circadian_strength	[0, 1]	Amplitude of 24h periodicity

References

• Goh, K.-I. & Barabási, A.-L. (2008). Burstiness and memory. EPL, 81(4).
• Hawkes, A.G. (1971). Self-exciting point processes. Biometrika, 58(1).

Structs

EventFeatures

Features extracted from event timestamps.

Enums

PointProcessError

Error types for point process analysis.

Functions

extract_event_features

Extract temporal point process features from a sequence of timestamps.