Enhanced time series predictability with well-defined structures

Huang, Yu; Fu, Zuntao

doi:10.1007/s00704-019-02836-6

Cited by 17 publications

(6 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The higher value of a with stronger short memory in PM 2.5 and PM 10 over southeast China corresponds to higher intrinsic predictability [44]. Studies also show that other meteorological variables over southeast China exhibit higher intrinsic predictability [45].…”

Section: Resultsmentioning

confidence: 88%

Memory Behaviors of Air Pollutions and Their Spatial Patterns in China

Nian

Qiao

et al. 2022

Front. Phys.

View full text Add to dashboard Cite

Particulate matter (PM2.5 and PM10) and ozone (O3) are the two major air pollutants in China in recent years. The fluctuations of PM2.5, PM10 and O3 strongly depend on the weather processes and anthropogenic emission. These processes may lead to the existence of short- and long-term memory behaviors in air pollutants. Hence, here we use the autoregressive parameter a of the first-order autoregressive process [AR (1)] to characterize the short-term memory effects of pollutants. We estimate the scaling exponent α using detrended fluctuation analysis (DFA) for the long-term memory effects of air pollutants (PM2.5, PM10, and O3) in summer and winter for different cities in China. Our results show that PM2.5, PM10, and O3 have strong short-term and long-term memory characteristics both in summer and winter. Furthermore, both the short- and long-term memory effects are stronger in winter than summer for most cities associated with stronger and longer persistent weather systems in winter. In general, the scaling exponent α of PM2.5 and PM10 are smaller for northern cities than those of southern cities in China. The long-term memory patterns of O3 are stronger in northern cities and weaker in southern cities in relative to those of PM2.5 and PM10 in winter. Our results show that the short- and long-term memory behaviors of air pollutions are dominated by the weather systems with different time scales.

show abstract

Section: Resultsmentioning

confidence: 88%

Memory Behaviors of Air Pollutions and Their Spatial Patterns in China

Nian

Qiao

et al. 2022

Front. Phys.

View full text Add to dashboard Cite

show abstract

“…As introduced in Section 3.2, SCINet learns an enhanced sequence representation that is fed into the fully-connected layer to yield the final prediction. To figure out what benefits the enhanced representation for prediction, inspired by [18,32], we utilize permutation entropy (PE) [6] to measure the predictability [26] of the original input and the enhanced representation learned by SCINet. Time series with lower PE values are thought less complex, thus theoretically easier to predict.…”

Section: Results and Analysismentioning

confidence: 99%

“…• We propose a hierarchical TSF framework, SCINet, based on the unique attributes of time series data. By iteratively extracting and exchanging information at different temporal resolutions, an effective representation with enhanced predictability can be learned, as verified by its comparatively lower permutation entropy (PE) [18].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Time Series is a Special Sequence: Forecasting with Sample Convolution and Interaction

Liu¹,

Zeng²,

Lai³

et al. 2021

Preprint

View full text Add to dashboard Cite

Time series is a special type of sequence data, a set of observations collected at even intervals of time and ordered chronologically. Existing deep learning techniques use generic sequence models (e.g., recurrent neural network, Transformer model, or temporal convolutional network) for time series analysis, which ignore some of its unique properties. For example, the downsampling of time series data often preserves most of the information in the data, while this is not true for general sequence data such as text sequence and DNA sequence. Motivated by the above, in this paper, we propose a novel neural network architecture and apply it for the time series forecasting problem, wherein we conduct sample convolution and interaction at multiple resolutions for temporal modeling. The proposed architecture, namely SCINet, facilitates extracting features with enhanced predictability. Experimental results show that SCINet achieves significant prediction accuracy improvement over existing solutions across various real-world time series forecasting datasets. In particular, it can achieve high forecasting accuracy for those temporal-spatial datasets without using sophisticated spatial modeling techniques. Our codes and data are presented in the supplemental material.Preprint. Under review.

show abstract

“…In order to fairly compare the errors of reconstructing different processes with different variability and units (Hyndman and Koehler, 2006;Pennekamp et al, 2019;Huang and Fu, 2019), we normalize the RMSE as…”

Section: Evaluation Of Reconstruction Qualitymentioning

confidence: 99%

Reconstructing coupled time series in climate systems using three kinds of machine-learning methods

Huang

Yang

2020

Earth Syst. Dynam.

Self Cite

View full text Add to dashboard Cite

Abstract. Despite the great success of machine learning, its application in climate dynamics has not been well developed. One concern might be how well the trained neural networks could learn a dynamical system and what will be the potential application of this kind of learning. In this paper, three machine-learning methods are used: reservoir computer (RC), backpropagation-based (BP) artificial neural network, and long short-term memory (LSTM) neural network. It shows that the coupling relations or dynamics among variables in linear or nonlinear systems can be inferred by RC and LSTM, which can be further applied to reconstruct one time series from the other. Specifically, we analyzed the climatic toy models to address two questions: (i) what factors significantly influence machine-learning reconstruction and (ii) how do we select suitable explanatory variables for machine-learning reconstruction. The results reveal that both linear and nonlinear coupling relations between variables do influence the reconstruction quality of machine learning. If there is a strong linear coupling between two variables, then the reconstruction can be bidirectional, and both of these two variables can be an explanatory variable for reconstructing the other. When the linear coupling among variables is absent but with the significant nonlinear coupling, the machine-learning reconstruction between two variables is direction dependent, and it may be only unidirectional. Then the convergent cross mapping (CCM) causality index is proposed to determine which variable can be taken as the reconstructed one and which as the explanatory variable. In a real-world example, the Pearson correlation between the average tropical surface air temperature (TSAT) and the average Northern Hemisphere SAT (NHSAT) is weak (0.08), but the CCM index of NHSAT cross mapped with TSAT is large (0.70). And this indicates that TSAT can be well reconstructed from NHSAT through machine learning. All results shown in this study could provide insights into machine-learning approaches for paleoclimate reconstruction, parameterization scheme, and prediction in related climate research.Highlights: i The coupling dynamics learned by machine learning can be used to reconstruct time series. ii Reconstruction quality is direction dependent and variable dependent for nonlinear systems. iii The CCM index is a potential indicator to choose reconstructed and explanatory variables. iv The tropical average SAT can be well reconstructed from the average Northern Hemisphere SAT.

show abstract

Enhanced time series predictability with well-defined structures

Cited by 17 publications

References 43 publications

Memory Behaviors of Air Pollutions and Their Spatial Patterns in China

Memory Behaviors of Air Pollutions and Their Spatial Patterns in China

Time Series is a Special Sequence: Forecasting with Sample Convolution and Interaction

Reconstructing coupled time series in climate systems using three kinds of machine-learning methods

Contact Info

Product

Resources

About