scholarly journals Estimation and classification of temporal trends to support integrated ecosystem assessment

2020 ◽  
Author(s):  
Hiroko Kato Solvang ◽  
Benjamin Planque
Keyword(s):  
Time Series ◽  
Barents Sea ◽  
Multivariate Time Series ◽  
Marine Ecosystem ◽  
Temporal Trends ◽  
Dynamic Factor ◽  
Short Time Series ◽  
Common Trends ◽  
Common Trend ◽  
Short Time

Abstract We propose a trend estimation and classification (TREC) approach to estimating dominant common trends among multivariate time series observations. Our methods are based on two statistical procedures that includes trend modelling and discriminant analysis for classifying similar trend (common trend) classes. We use simulations to evaluate the proposed approach and compare it with a relevant dynamic factor analysis in the time domain, which was recently proposed to estimate common trends in fisheries time series. We apply the TREC approach to the multivariate short time series datasets investigated by the ICES integrated assessment working groups for the Norwegian Sea and the Barents Sea. The proposed approach is robust for application to short time series, and it directly identifies and classifies the dominant trends underlying observations. Based on the classified trend classes, we suggest that communication among stakeholders like marine managers, industry representatives, non-governmental organizations, and governmental agencies can be enhanced by finding the common tendency between a biological community in a marine ecosystem and the environmental factors, as well as by the icons produced by generalizing common trend patterns.

scholarly journals Estimating common trends in multivariate time series using dynamic factor analysis

2003 ◽  
Vol 14 (7) ◽  
pp. 665-685 ◽  
Author(s):  
A. F. Zuur ◽  
R. J. Fryer ◽  
I. T. Jolliffe ◽  
R. Dekker ◽  
J. J. Beukema
Keyword(s):  
Time Series ◽  
Factor Analysis ◽  
Multivariate Time Series ◽  
Dynamic Factor Analysis ◽  
Dynamic Factor ◽  
Common Trends

scholarly journals Deep learning for clustering of multivariate clinical patient trajectories with missing values

GigaScience ◽  
2019 ◽  
Vol 8 (11) ◽  
Author(s):  
Johann de Jong ◽  
Mohammad Asif Emon ◽  
Ping Wu ◽  
Reagon Karki ◽  
Meemansa Sood ◽  
...  
Keyword(s):  
Time Series ◽  
Deep Learning ◽  
Alzheimer Disease ◽  
Parkinson Disease ◽  
Disease Progression ◽  
Missing Values ◽  
Multivariate Time Series ◽  
Short Time Series ◽  
Deep Embedding ◽  
Short Time

Abstract Background Precision medicine requires a stratification of patients by disease presentation that is sufficiently informative to allow for selecting treatments on a per-patient basis. For many diseases, such as neurological disorders, this stratification problem translates into a complex problem of clustering multivariate and relatively short time series because (i) these diseases are multifactorial and not well described by single clinical outcome variables and (ii) disease progression needs to be monitored over time. Additionally, clinical data often additionally are hindered by the presence of many missing values, further complicating any clustering attempts. Findings The problem of clustering multivariate short time series with many missing values is generally not well addressed in the literature. In this work, we propose a deep learning–based method to address this issue, variational deep embedding with recurrence (VaDER). VaDER relies on a Gaussian mixture variational autoencoder framework, which is further extended to (i) model multivariate time series and (ii) directly deal with missing values. We validated VaDER by accurately recovering clusters from simulated and benchmark data with known ground truth clustering, while varying the degree of missingness. We then used VaDER to successfully stratify patients with Alzheimer disease and patients with Parkinson disease into subgroups characterized by clinically divergent disease progression profiles. Additional analyses demonstrated that these clinical differences reflected known underlying aspects of Alzheimer disease and Parkinson disease. Conclusions We believe our results show that VaDER can be of great value for future efforts in patient stratification, and multivariate time-series clustering in general.

Estimators of Wave Peak Period for Large Return Period Seastates

2011 ◽  
Author(s):  
Michele Drago ◽  
Tiziana Ciuffardi ◽  
Giancarlo Giovanetti
Keyword(s):  
Time Series ◽  
Return Period ◽  
Barents Sea ◽  
Western Mediterranean ◽  
Tyrrhenian Sea ◽  
Peak Period ◽  
Short Time Series ◽  
Functional Relationships ◽  
Measured Time ◽  
Short Time

Different functional relationships for the distribution of the peak period Tp for a given seastate Hs are tested against hindcasted and measured time series for various locations in the European seas, namely Barents Sea, Baltic Sea and Western Mediterranean (Tyrrhenian) Sea. The aim is to investigate their performance when extrapolating from available data for the estimation of the long return period seastates which are not covered by the length of the time series. A second complementary objective of the paper is to quantify the uncertainty and the variability that can occur in the estimation of the Hs/Tp association when a short time series is available. This is done by assessing the Hs/Tp relation from a 35-years long time series that will be considered as the correct one. Then it will be compared with those obtained by analysing multiple cases of shorter and shorter portions of the time series. The spread of the results against the one obtained with the entire time series will be estimated to give some indications about how long a time series has to be to ensure a correct extrapolation to the long return period seastates and to provide an indication of the error that can affect the results when a short time series is available.

scholarly journals Finding the direction of lowest resilience in multivariate complex systems

2019 ◽  
Vol 16 (159) ◽  
pp. 20190629 ◽  
Author(s):  
Els Weinans ◽  
J. Jelle Lever ◽  
Sebastian Bathiany ◽  
Rick Quax ◽  
Jordi Bascompte ◽  
...  
Keyword(s):  
Time Series ◽  
Complex Systems ◽  
Financial Markets ◽  
Multivariate Time Series ◽  
Network Systems ◽  
Short Time Series ◽  
Novel Approach ◽  
Data Resolution ◽  
Multivariate Systems ◽  
Short Time

The dynamics of complex systems, such as ecosystems, financial markets and the human brain, emerge from the interactions of numerous components. We often lack the knowledge to build reliable models for the behaviour of such network systems. This makes it difficult to predict potential instabilities. We show that one could use the natural fluctuations in multivariate time series to reveal network regions with particularly slow dynamics. The multidimensional slowness points to the direction of minimal resilience, in the sense that simultaneous perturbations on this set of nodes will take longest to recover. We compare an autocorrelation-based method with a variance-based method for different time-series lengths, data resolution and different noise regimes. We show that the autocorrelation-based method is less robust for short time series or time series with a low resolution but more robust for varying noise levels. This novel approach may help to identify unstable regions of multivariate systems or to distinguish safe from unsafe perturbations.

scholarly journals Identifying bidirectional total and non-linear information flow in functional corticomuscular coupling during a dorsiflexion task: a pilot study

2021 ◽  
Vol 18 (1) ◽  
Author(s):  
Tie Liang ◽  
Qingyu Zhang ◽  
Xiaoguang Liu ◽  
Bin Dong ◽  
Xiuling Liu ◽  
...  
Keyword(s):  
Time Series ◽  
Information Flow ◽  
Gamma Band ◽  
Motor Dysfunction ◽  
Beta Band ◽  
Short Time Series ◽  
Information Interaction ◽  
Stroke Group ◽  
Short Time ◽  
Maximal Information Coefficient

Abstract Background The key challenge to constructing functional corticomuscular coupling (FCMC) is to accurately identify the direction and strength of the information flow between scalp electroencephalography (EEG) and surface electromyography (SEMG). Traditional TE and TDMI methods have difficulty in identifying the information interaction for short time series as they tend to rely on long and stable data, so we propose a time-delayed maximal information coefficient (TDMIC) method. With this method, we aim to investigate the directional specificity of bidirectional total and nonlinear information flow on FCMC, and to explore the neural mechanisms underlying motor dysfunction in stroke patients. Methods We introduced a time-delayed parameter in the maximal information coefficient to capture the direction of information interaction between two time series. We employed the linear and non-linear system model based on short data to verify the validity of our algorithm. We then used the TDMIC method to study the characteristics of total and nonlinear information flow in FCMC during a dorsiflexion task for healthy controls and stroke patients. Results The simulation results showed that the TDMIC method can better detect the direction of information interaction compared with TE and TDMI methods. For healthy controls, the beta band (14–30 Hz) had higher information flow in FCMC than the gamma band (31–45 Hz). Furthermore, the beta-band total and nonlinear information flow in the descending direction (EEG to EMG) was significantly higher than that in the ascending direction (EMG to EEG), whereas in the gamma band the ascending direction had significantly higher information flow than the descending direction. Additionally, we found that the strong bidirectional information flow mainly acted on Cz, C3, CP3, P3 and CPz. Compared to controls, both the beta-and gamma-band bidirectional total and nonlinear information flows of the stroke group were significantly weaker. There is no significant difference in the direction of beta- and gamma-band information flow in stroke group. Conclusions The proposed method could effectively identify the information interaction between short time series. According to our experiment, the beta band mainly passes downward motor control information while the gamma band features upward sensory feedback information delivery. Our observation demonstrate that the center and contralateral sensorimotor cortex play a major role in lower limb motor control. The study further demonstrates that brain damage caused by stroke disrupts the bidirectional information interaction between cortex and effector muscles in the sensorimotor system, leading to motor dysfunction.

scholarly journals A reliable measure of similarity based on dependency for short time series: an application to gene expression networks

2009 ◽  
Vol 10 (1) ◽  
pp. 270 ◽  
Author(s):  
Mônica G Campiteli ◽  
Frederico M Soriani ◽  
Iran Malavazi ◽  
Osame Kinouchi ◽  
Carlos AB Pereira ◽  
...  
Keyword(s):  
Gene Expression ◽  
Time Series ◽  
Reliable Measure ◽  
Short Time Series ◽  
Measure Of Similarity ◽  
Short Time

scholarly journals LIMITATIONS OF THE PANEL REGRESSION MODEL APPLICATION: THE EXAMPLE OF THE WESTERN BALKAN COUNTRIES

2021 ◽  
Vol 18 (32) ◽  
Author(s):  
Stanko Stanić ◽  
Bojan Baškot
Keyword(s):  
Time Series ◽  
Regression Model ◽  
Time Frame ◽  
Panel Regression ◽  
Time Dimension ◽  
Data Set ◽  
Short Time Series ◽  
Western Balkan Region ◽  
Balkan Region ◽  
Short Time

Panel regression model may seem like an appealing solution in conditions of limited time series. This is often used as a shortcut to achieve deeper data set by setting several individual cases on the same time dimension, where cross units visually but not really multiply a time frame. Macroeconometrics of the Western Balkan region assumes short time series issue. Additionally, the structural brakes are numerous. Panel regression may seem like a solution, but there are some limitations that should be considered.

Sign in / Sign up

Export Citation Format

Share Document