A wavelet-based approach for imputation in nonstationary multivariate time series

Wilson, Rebecca E.; Eckley, Idris A.; Nunes, Matthew; Park, Timothy Alexander

doi:10.1007/s11222-021-09998-2

Cited by 11 publications

(3 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The above signifies that a missing-values prediction method based on spectral analysis would most likely make accurate predictions for most of the data variance, but it may not account for any random irregularities, such as extreme meteorological events, other anthropogenic factors (e.g., marine traffic), or events related to earth dynamics (i.e., earthquakes and sediment dynamics). Missing value imputation methods based on the frequency domain are either using the Fourier [55][56][57] or the wavelet (e.g., in [58]) spectrum of the time series.…”

Section: Spectral Analysis With Incomplete Data: the Clean Algorithmmentioning

confidence: 99%

Using Distributed Temperature Sensing for Long-Term Monitoring of Pockmark Activity in the Gulf of Patras (Greece): Data Processing Hints and Preliminary Findings

Fakiris,

Papatheodorou,

Christodoulou

et al. 2023

Sensors

View full text Add to dashboard Cite

Patras Gulf pockmark field (Western Greece) is a tectonically controlled field that has been activated at least twice by strong earthquakes (M5.4, 14 July 1993 and M6.4, 8 June 2008), and episodic gas seepages have been recorded in the past using geophysical means. A distributed temperature sensor (DTS) system was deployed inside a shallow pockmark and along an active fault at the northern end of the field. This ongoing experiment represents the first long-term monitoring ever conducted on gas-bearing pockmarks and active faults by the DTS system. For now, we have acquired and analyzed data regarding about 1.56 years. One of the primary objectives of this study is to establish methodological queues for data processing and analysis, including spectral analysis and incomplete data treatment techniques, to be standardized for use in further stages of the experiment. Spectral analysis was proven capable of separating the temperature footprint of background environmental components, such as sea-atmosphere heat flux, tides, and winds/waves, from high-frequency temperature residuals. Those residuals represent unusual events that might be correlated to seismicity. Monitoring the causal relationship between seismic activity and seabed water temperature changes in the field was thus attempted. No significant local earthquakes occurred during the monitoring period. Although the relation between seismicity and irregular seabed water temperature events was not systematic, we postulate that four thermal events have a causative link with the local seismicity. The DTS system constitutes a low-cost monitoring system, and the promising preliminary results of this experiment suggest that it is worth testing for a longer period.

show abstract

Section: Spectral Analysis With Incomplete Data: the Clean Algorithmmentioning

confidence: 99%

Using Distributed Temperature Sensing for Long-Term Monitoring of Pockmark Activity in the Gulf of Patras (Greece): Data Processing Hints and Preliminary Findings

Fakiris,

Papatheodorou,

Christodoulou

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Wavelet decomposition can also fill in missing values in time series [17][18][19]. In this study, 3300 uplink traffic data with no missing value were selected as the real data in the cell No.…”

Section: Comparison Of Tensor Decomposition and Wavelet Decompositionmentioning

confidence: 99%

Short term prediction of wireless traffic based on tensor decomposition and recurrent neural network

et al. 2021

View full text Add to dashboard Cite

This paper proposes a wireless network traffic prediction model based on Bayesian Gaussian tensor decomposition and recurrent neural network with rectified linear unit (BGCP-RNN-ReLU model), which can effectively predict the changes in the upstream and downstream network traffic in a short period of time in the future. The research is divided into two parts: (i) The missing observations are imputed by an algorithm based on Bayesian Gaussian tensor decomposition. (ii) The recurrent neural network is used to forecast the true observations only rather than both true and estimated observations. The results show that, compared with other combined models of missing data imputation and neural networks, the BGCP-RNN-ReLU model proposed in this paper has the smallest prediction error for both the upstream and downstream traffic. The new model achieves better forecasting precision, and thus can help to regulate the load of communication station to reduce resource consumption. Highlights The problem of forecasting wireless network traffic with missing values is divided in two stages to handle. A newly propose d method can more efficiently impute missing values in wireless network traffic data. Simple recurrent neural network obtains better prediction performance than other complex networks.

show abstract

“…Existing methods for analyzing irregular time series can be categorized into three main directions [9]: (i) the repair approach in which missing observations are recovered via smoothing or imputation [10][11][12][13][14]-also implemented, especially in recent years, by machine learning methods [15][16][17][18]; (ii) the generalization of spectral analysis tools [19,20], such as wavelets [21][22][23][24]; (iii) kernel methods [25,26]. In this paper, we deal with a repair approach which uses an input preparation step based on machine learning.…”

Section: Introductionmentioning

confidence: 99%

Ensemble Methods for Jump-Diffusion Models of Power Prices

Mari

Baldassari

2021

Energies

View full text Add to dashboard Cite

We propose a machine learning-based methodology which makes use of ensemble methods with the aims (i) of treating missing data in time series with irregular observation times and detecting anomalies in the observed time behavior; (ii) of defining suitable models of the system dynamics. We applied this methodology to US wholesale electricity price time series that are characterized by missing data, high and stochastic volatility, jumps and pronounced spikes. For missing data, we provide a repair approach based on the missForest algorithm, an imputation algorithm which is completely agnostic about the data distribution. To identify anomalies, i.e., turbulent movements of power prices in which jumps and spikes are observed, we took into account the no-gap reconstructed electricity price time series, and then we detected anomalous regions using the isolation forest algorithm, an anomaly detection method that isolates anomalies instead of profiling normal data points as in the most common techniques. After removing anomalies, the additional gaps will be newly filled by the missForest imputation algorithm. In this way, a complete and clean time series describing the stable dynamics of power prices can be obtained. The decoupling between the stable motion and the turbulent motion allows us to define suitable jump-diffusion models of power prices and to provide an estimation procedure that uses the full information contained in both the stable and the turbulent dynamics.

show abstract

A wavelet-based approach for imputation in nonstationary multivariate time series

Cited by 11 publications

References 69 publications

Using Distributed Temperature Sensing for Long-Term Monitoring of Pockmark Activity in the Gulf of Patras (Greece): Data Processing Hints and Preliminary Findings

Using Distributed Temperature Sensing for Long-Term Monitoring of Pockmark Activity in the Gulf of Patras (Greece): Data Processing Hints and Preliminary Findings

Short term prediction of wireless traffic based on tensor decomposition and recurrent neural network

Ensemble Methods for Jump-Diffusion Models of Power Prices

Contact Info

Product

Resources

About