# Empirical Identification of Non-stationary Dynamics in Time Series of Recordings

This source preferred by Emili Balaguer-Ballester

**Authors: **Balaguer-Ballester, E., Tabas-Diaz, A. and Budka, M.

**Volume:** 8779 LNAI

**Pages:** 142-151

**Publisher:** Elsevier

**DOI:** 10.1007/978-3-319-11298-5-15

Non-stationarity time series are very common in physical, biological and in real-world systems in general, ranging from geophysics, econometrics or electroencephalography to logistics. Identifying, detecting and adapting learning algorithms to non-stationary environments is a fundamental task in many data mining scenarios; however it is often a major challenge for current methodologies. Data analysis in the context of time-varying statistical moments is a very active research direction in machine learning and in computational statistics; but theoretical insights into latent causes of non-stationarity in empirical data are very scarce. In this study, we evaluate the capacity of the trajectory classification error statistic in order to detect a significant variation in the underlying dynamics of data collected in multiple stages. We analysed qualitatively the conditions leading to observable changes in non-stationary data generated by Duffing non-linear oscillators; which are ubiquitous models of complex classification problems. Analyses are further benchmarked in a dataset consisting of atmospheric pollutants time series. © 2014 Springer International Publishing Switzerland.

This source preferred by Marcin Budka

This data was imported from Scopus:

**Authors: **Balaguer-Ballester, E., Tabas-Diaz, A. and Budka, M.

**Volume:** 8779 LNAI

**Pages:** 142-151

**Publisher:** Springer Verlag

**ISBN:** 9783319112978

**DOI:** 10.1007/978-3-319-11298-5-15

Non-stationarity time series are very common in physical, biological and in real-world systems in general, ranging from geophysics, econometrics or electroencephalography to logistics. Identifying, detecting and adapting learning algorithms to non-stationary environments is a fundamental task in many data mining scenarios; however it is often a major challenge for current methodologies. Data analysis in the context of time-varying statistical moments is a very active research direction in machine learning and in computational statistics; but theoretical insights into latent causes of non-stationarity in empirical data are very scarce. In this study, we evaluate the capacity of the trajectory classification error statistic in order to detect a significant variation in the underlying dynamics of data collected in multiple stages. We analysed qualitatively the conditions leading to observable changes in non-stationary data generated by Duffing non-linear oscillators; which are ubiquitous models of complex classification problems. Analyses are further benchmarked in a dataset consisting of atmospheric pollutants time series. © 2014 Springer International Publishing Switzerland.

This data was imported from Web of Science (Lite):

**Authors: **Balaguer-Ballester, E., Tabas-Diaz, A. and Budka, M.

**Volume:** 8779

**Pages:** 142-151

**ISBN:** 978-3-319-11297-8