Pseudo Periodic Synthetic Time Series

Donated on 2/7/1999

This data set is designed for testing indexing schemes in time series databases. The data appears highly periodic, but never exactly repeats itself.

Dataset Characteristics

Univariate, Time-Series

Subject Area

Other

Associated Tasks

-

Feature Type

-

# Instances

100000

# Features

-

Dataset Information

Additional Information

This data set is designed for testing indexing schemes in time series databases. It is a much larger dataset than has been used in any published study (That we are currently aware of). It contains one million data points. The data has been split into 10 sections to facilitate testing (see below). We recommend building the index with 9 of the 100,000-datapoint sections, and randomly extracting a query shape from the 10th section. (Some previously published work seems to have used queries that were also used to build the indexing structure. This will produce optimistic results) The data are interesting because they have structure at different resolutions. Each of the 10 sections where generated by independent invocations of the function: (see equation.gif) Where rand(x) produces a random integer between zero and x. The data appears highly periodic, but never exactly repeats itself. This feature is designed to challenge the indexing structure. The time series are ploted here: (ts1-5.gif), (ts6-10.gif)

Has Missing Values?

No

Variable Information

The data is stored in one ASCII file. There are 10 columns, 100,000 rows. All data points are in the range -0.5 to +0.5. Rows are separated by carriage returns, columns by spaces.

Reviews

There are no reviews for this dataset yet.

Login to Write a Review
Download
0 citations
2119 views

Creators

Eamonn Keogh

Michael Pazzani

License

By using the UCI Machine Learning Repository, you acknowledge and accept the cookies and privacy practices used by the UCI Machine Learning Repository.

Read Policy