-
Maximum likelihood estimation of burst-merging kernels for bursty time series
Authors:
Tibebe Birhanu,
Hang-Hyun Jo
Abstract:
Various time series in natural and social processes have been found to be bursty. Events in the time series rapidly occur within short time periods, forming bursts, which are alternated with long inactive periods. As the timescale defining bursts increases, individual events are sequentially merged to become small bursts and then bigger ones, eventually leading to the single burst containing all e…
▽ More
Various time series in natural and social processes have been found to be bursty. Events in the time series rapidly occur within short time periods, forming bursts, which are alternated with long inactive periods. As the timescale defining bursts increases, individual events are sequentially merged to become small bursts and then bigger ones, eventually leading to the single burst containing all events. Such a merging pattern has been depicted by a tree that fully reveals the hierarchical structure of bursts, thus called a burst tree. The burst-tree structure can be simply characterized by a burst-merging kernel that dictates which bursts are merged together as the timescale increases. In this work, we develop the maximum likelihood estimation method of the burst-merging kernel from time series, which is successfully tested against the time series generated using several model kernels. We also apply our method to some empirical time series from various backgrounds. Our method provides a useful tool to precisely characterize the time series data, hence enabling to study their underlying mechanisms more accurately.
△ Less
Submitted 18 March, 2025;
originally announced March 2025.
-
Burst-tree structure and higher-order temporal correlations
Authors:
Tibebe Birhanu,
Hang-Hyun Jo
Abstract:
Understanding characteristics of temporal correlations in time series is crucial for developing accurate models in natural and social sciences. The burst-tree decomposition method was recently introduced to reveal higher-order temporal correlations in time series in a form of an event sequence, in particular, the hierarchical structure of bursty trains of events for the entire range of timescales…
▽ More
Understanding characteristics of temporal correlations in time series is crucial for developing accurate models in natural and social sciences. The burst-tree decomposition method was recently introduced to reveal higher-order temporal correlations in time series in a form of an event sequence, in particular, the hierarchical structure of bursty trains of events for the entire range of timescales [Jo et al., Sci.~Rep.~\textbf{10}, 12202 (2020)]. Such structure has been found to be simply characterized by the burst-merging kernel governing which bursts are merged together as the timescale for defining bursts increases. In this work, we study the effects of kernels on the higher-order temporal correlations in terms of burst size distributions, memory coefficients for bursts, and the autocorrelation function. We employ several kernels, including the constant, sum, product, and diagonal kernels as well as those inspired by empirical results. We generically find that kernels with preferential merging lead to heavy-tailed burst size distributions, while kernels with assortative merging lead to positive correlations between burst sizes. The decaying exponent of the autocorrelation function depends not only on the kernel but also on the power-law exponent of the interevent time distribution. In addition, thanks to the analogy to the coagulation process, analytical solutions of burst size distributions for some kernels could be obtained. Our findings may shed light on the role of burst-merging kernels as underlying mechanisms of higher-order temporal correlations in time series.
△ Less
Submitted 9 December, 2024; v1 submitted 3 September, 2024;
originally announced September 2024.
-
Temporal scaling theory for bursty time series with clusters of arbitrarily many events
Authors:
Hang-Hyun Jo,
Tibebe Birhanu,
Naoki Masuda
Abstract:
Long-term temporal correlations in time series in a form of an event sequence have been characterized using an autocorrelation function (ACF) that often shows a power-law decaying behavior. Such scaling behavior has been mainly accounted for by the heavy-tailed distribution of interevent times (IETs), i.e., the time interval between two consecutive events. Yet little is known about how correlation…
▽ More
Long-term temporal correlations in time series in a form of an event sequence have been characterized using an autocorrelation function (ACF) that often shows a power-law decaying behavior. Such scaling behavior has been mainly accounted for by the heavy-tailed distribution of interevent times (IETs), i.e., the time interval between two consecutive events. Yet little is known about how correlations between consecutive IETs systematically affect the decaying behavior of the ACF. Empirical distributions of the burst size, which is the number of events in a cluster of events occurring in a short time window, often show heavy tails, implying that arbitrarily many consecutive IETs may be correlated with each other. In the present study, we propose a model for generating a time series with arbitrary functional forms of IET and burst size distributions. Then, we analytically derive the ACF for the model time series. In particular, by assuming that the IET and burst size are power-law distributed, we derive scaling relations between power-law exponents of the ACF decay, IET distribution, and burst size distribution. These analytical results are confirmed by numerical simulations. Our approach helps to rigorously and analytically understand the effects of correlations between arbitrarily many consecutive IETs on the decaying behavior of the ACF.
△ Less
Submitted 25 April, 2024;
originally announced April 2024.
