Explaining Interval Sequences by Randomization

Abstract

Sequences of events are an ubiquitous form of data. In this paper, we show that it is feasible to present an event sequence as an interval sequence. We show how sequences can be efficiently randomized, how to choose a correct null model and how to use randomizations to derive confidence intervals. Using these techniques, we gain knowledge of the temporal structure of the sequence. Time and Fourier space representations, autocorrelations and arbitrary features can be used as constraints in investigating the data. The methods presented are applied to two real-life datasets; a medical heart interbeat interval dataset and a word dataset from a book. We find that the interval sequence representation and randomization methods provide a powerful way to explore interval sequences and explain their structure.

Cite

Text

Henelius et al. "Explaining Interval Sequences by Randomization." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2013. doi:10.1007/978-3-642-40988-2_22

Markdown

[Henelius et al. "Explaining Interval Sequences by Randomization." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2013.](https://mlanthology.org/ecmlpkdd/2013/henelius2013ecmlpkdd-explaining/) doi:10.1007/978-3-642-40988-2_22

BibTeX

@inproceedings{henelius2013ecmlpkdd-explaining,
  title     = {{Explaining Interval Sequences by Randomization}},
  author    = {Henelius, Andreas and Korpela, Jussi and Puolamäki, Kai},
  booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
  year      = {2013},
  pages     = {337-352},
  doi       = {10.1007/978-3-642-40988-2_22},
  url       = {https://mlanthology.org/ecmlpkdd/2013/henelius2013ecmlpkdd-explaining/}
}