Abstract:Frequent periodic pattern mining has been studied a lot in the last decade, with various algorithms already proposed for mining different kinds of periodic patterns, from event sequence, spatiotemporal data, etc.. In this paper, instead of using the Maximalsubpattern Tree structure adopted widely in previous algorithms, we proposed a new algorithm under pattern-growth framework which was widely used in generalized sequential pattern mining. Based on that, we further proposed pruning/checking strategies for mining maximal periodic patterns. We further proposed a hybrid method combining the strength of both the Maximal-subpattern Tree and our proposed maximal pattern mining algorithm, to discover maximal periodic patterns more efficiently. Experimental results show that comparing to existing algorithm MTHS, our frequent periodic pattern mining algorithm is thousands times faster, and the devised maximal periodic pattern mining algorithm is even more than hundreds times faster than the former proposed frequent periodic pattern mining algorithm on some datasets. Comparing to both the two proposed frequent (maximal) pattern mining algorithms, the novel hybrid method for mining maximal patterns shows even greater performance advantage, with the running time further reduced for thousands times. Finally, we proposed a strategy for mining common frequent (maximal) periodic patterns and gave a novel case study on the discovery of common animal migration behaviors from a large real GPS-tracking dataset. Some of the discovered patterns are very interesting, like turkey vultures move to Mexico in Winter, while get back to US in Summer.

Mining asynchronous periodic patterns in time series data

Mining Asynchronous Periodic Patterns in Time Series Data.

Periodicity Detection Method of Periodic Pattern Mining in Time Series

An Asynchronous Periodic Sequential Patterns Mining Algorithm with Multiple Minimum Item Supports.

An Asynchronous Periodic Sequential Pattern Mining Algorithm with Multiple Minimum Item Supports for Ad Hoc Networking

Efficient Mining of Event Periodicity in Data Series

Mining Surprising Periodic Patterns

Granular-based Partial Periodic Pattern Discovery over Time Series Data

Segmental semi-markov model based online series pattern detection under arbitrary time scaling

Infominer: mining surprising periodic patterns

Efficient Discovery of Periodic Patterns over Event Sequences

STAMP: On Discovery of Statistically Important Pattern Repeats in Long Sequential Data

Periodicity-Oriented Data Analytics on Time-Series Data for Intelligence System

Mining Sequential Patterns with Periodic General Gap Constraints

A Two-Phase Approach for Unexpected Pattern Mining.

Online Series Pattern Detection Based on Advanced Segmental Semi-Markov Model

InfoMiner+: Mining Partial Periodic Patterns with Gap Penalties.

Discovering Periodic Patterns Common to Multiple Sequences

Fuzzy-driven periodic frequent pattern mining

Robust Dominant Periodicity Detection for Time Series with Missing Data

A Hybrid Online Series Pattern Detection Algorithm