On the Discovery of Weak Periodicities in Large Time Series

Get BibTex-formatted data

Download

PDF

Author

Christos Berberidis, Ioannis Vlahavas, Walid G. Aref, Mikhail Atallah, Ahmed Elmagarmid

Tech report number

CERIAS TR 2003-17

Entry type

inproceedings

Abstract

The search for weak periodic signals in time series data is an active topic of research. Given the fact that rarely a real world dataset is perfectly periodic, this paper approaches this problem in terms of data mining, trying to discover approximate and partial periodicities, when no period length is known in advance. We utilize the autocorrelation function in order to extract partial periodicities from large time series. In existing time series mining algorithms, the period length is user-specified. We propose an algorithm that extracts a set of candidate periods featured in a time series that satisfy a minimum confidence threshold, by utilizing the autocorrelation function and FFT as a filter. We extend this technique for capturing approximate periodicities. We provide some mathematical background as well as experimental results.

Download

PDF

Date

2002 – 08

Booktitle

6th European Conference on Principles and Practice of Knowledge Discovery in Databases

Key alpha

Atallah

Pages

51-61

Affiliation

Aristotle University of Thessaloniki, Purdue University

Publication Date

2002-08-01

Language

English

Location

A hard-copy of this is in the CERIAS Library

BibTex-formatted data

To refer to this entry, you may select and copy the text below and paste it into your BibTex document. Note that the text may not contain all macros that BibTex supports.

@Inproceedings{ Atallah,
	title = "On the Discovery of Weak Periodicities in Large Time Series",
	author = "Christos Berberidis, Ioannis Vlahavas, Walid G. Aref, Mikhail Atallah, Ahmed Elmagarmid",
	year = "2002",
	month = "08",
	booktitle = "6th European Conference on Principles and Practice of Knowledge Discovery in Databases",
	pages = "51-61",
	abstract = "The search for weak periodic signals in time series data is an active topic of research. Given the fact that rarely a real world dataset is perfectly periodic, this paper approaches this problem in terms of data mining, trying to discover approximate and partial periodicities, when no period length is known in advance.  We utilize the autocorrelation function in order to extract partial periodicities from large time series.  In existing time series mining algorithms, the period length is user-specified.  We propose an algorithm that extracts a set of candidate periods featured in a time series that satisfy a minimum confidence threshold, by utilizing the autocorrelation function and FFT as a filter.  We extend this technique for capturing approximate periodicities.  We provide some mathematical background as well as experimental results.",
	affiliation = "Aristotle University of Thessaloniki, Purdue University",
	language = "English",
}