Representation Of Time Series Research Articles

Comment on “Five centuries of cod catches in eastern Canada,” by Schijns et al

Abstract Schijns et al. use a historical time series to inform a stock assessment model for the northern Atlantic cod (Gadus morhua) fishery from 1508 to 2019. They find that catches from the sixteenth century to the 1950s did not exceed 200000 t per annum and could have been sustained today “if fishing effort and mortality had been stabilized in the 1980s”. Had Schijns et al. used a more complete and representative time series (as identified below), they would have found that catches were substantially higher during much of the time period, possibly affecting their conclusions regarding the timing and onset of unsustainable exploitation. In an earlier paper, based on original archival documentation, we have argued that total landings in the Newfoundland fishery averaged 400000 t in the eighteenth century and peaked at 600000 t in 1788. We contend that pre-industrial technology was sufficient to have a significant impact on marine life.

Reweighting the OHS and GHS to improve data quality: Representativeness, household counts, and small households

AbstractThe October Household Surveys (OHS) (1994–1999) and the General Household Surveys (GHS) (2002–present) collected by StatsSA comprise South Africa's only nationally representative time series with information on both people and households for (almost) every year of the post‐apartheid period. However, the quality of these data has been compromised by how the survey weights have been calibrated. We document these problems and their implications in detail and then use cross‐entropy estimation to recalibrate the survey weights for a stacked version of these surveys between 1995 and 2011 to address these weaknesses. The first issue with the weights is that calibration procedure breaks with sampling practise by calibrating person and household weights separately. This creates conceptual problems because the data are not properly representative of the population. It also creates statistical problems, including that a series of total population and household counts cannot be reliably extracted from the series, which is typically a first‐order output for such a time series. Second, the series of household counts extracted from the GHS is probably too low. Third, no compensation is made by the survey weights for the chronic undersampling of small households over the entire period. Our new weights make headway in resolving these issues.

AvgNet: Adaptive Visibility Graph Neural Network and Its Application in Modulation Classification

Our digital world is full of time series and graphs which capture the various aspects of many complex systems. Traditionally, there are respective methods in processing these two different types of data, e.g., Recurrent Neural Network (RNN) and Graph Neural Network (GNN), while in recent years, time series could be mapped to graphs by using the techniques such as Visibility Graph (VG), which simultaneously captures relevant aspects of both local and global dynamics in an easy way, so that researchers can use graph algorithms to mine the knowledge in time series and gain special latent graph representation features. Such mapping methods establish a bridge between time series and graphs, and have high potential to facilitate the analysis of various real-world time series. However, the VG method and its variants are just based on fixed rules and thus lack of flexibility, largely limiting their application in reality. In this paper, we propose an Adaptive Visibility Graph (AVG) algorithm that can adaptively map time series into graphs, based on which we further establish an end-to-end classification framework AvgNet, by utilizing GNN model DiffPool as the classifier. We then adopt AvgNet for radio signal modulation classification which is an important task in the field of wireless communication. The simulations validate that AvgNet outperforms a series of advanced deep learning methods, achieving the state-of-the-art performance in this task.

Electric demand forecasting with neural networks and symbolic time series representations

This paper addresses the electric demand prediction problem using neural networks and symbolization techniques. Symbolization techniques provide a time series symbolic representation of a lower length than the original time series. In our methodology, we incorporate the use of encoding from ordinal regression, preserving the notation of order between the symbols and make extensive experimentation with different neural network architectures and symbolization techniques. In our experimentation, we used the total electric demand data in the Spanish peninsula electric network, taken from 2009 to 2019 with a granularity of 10 min. The best model found making use of the symbolization methodology offered us slightly worse quality metrics (1.3655 RMSE and 0.0390 MAPE instead of the 1.2889 RMSE and 0.0363 MAPE from the best numerical model) but it was trained 6826 times faster.

Bayesian imputation of COVID-19 positive test counts for nowcasting under reporting lag.

Obtaining up to date information on the number of UK COVID‐19 regional infections is hampered by the reporting lag in positive test results for people with COVID‐19 symptoms. In the UK, for ‘Pillar 2’ swab tests for those showing symptoms, it can take up to five days for results to be collated. We make use of the stability of the under reporting process over time to motivate a statistical temporal model that infers the final total count given the partial count information as it arrives. We adopt a Bayesian approach that provides for subjective priors on parameters and a hierarchical structure for an underlying latent intensity process for the infection counts. This results in a smoothed time‐series representation nowcasting the expected number of daily counts of positive tests with uncertainty bands that can be used to aid decision making. Inference is performed using sequential Monte Carlo.

And we thought the Millennium Drought was bad: Assessing climate variability and change impacts on an Australian dryland wetland using an ecohydrologic emulator

During the Millennium Drought in southeast Australia (2001–2009), dryland wetlands experienced widespread ecological deterioration, which highlighted their vulnerability to natural climate variability and the potential effects of drying climate change. Here we use 30-year observed streamflow data (1991–2020) and numerical models to assess the impacts of climate variability and climate change on the Macquarie Marshes (the Marshes), a large floodplain wetland complex in the semi-arid region of New South Wales, Australia. A fast ecohydrologic emulator based on network linear programming with side constraints was developed to simulate the spatial and temporal responses of different wetland vegetation types to water regime. The emulator represents the wetland by a series of inter-connected level-pool reservoirs with the volume-discharge relationship obtained from a calibrated quasi-2d hydrodynamic model. The emulator reproduces daily flows and volume with good accuracy (Nash-Sutcliffe statistic ranging from 0.61 to 0.96) with 1/26,000 of the computational effort. We use the emulator to simulate the potential effects of climatic variability on vegetation by running the model over 30 years of observed data and 1000 statistically representative 30-year streamflow time series, which were generated using a stochastic model calibrated to the gauged flows. The collection of results for all 1000 contemporary simulations indicates the Marshes experience severe conditions 43% (± 18%) of the time in a 30-year period. We then ran an additional 6000 simulations to assess the combined impacts of climate variability and future climate change at the end of the century. For the driest future climates (-60% and -30% reduction in runoff), the Marshes remain in severe condition 89% (± 6%) and 63% (± 16%) of the time, respectively, while no major differences with respect to the contemporary conditions were found for the wetter future. Our results highlight the importance of quantifying the extent and uncertainty in the degradation of these ecosystems due to climate variability and change for informing management decisions.

Interkulturelle Musiktherapie –Ein Literaturreview

Cultural diversity has become a fact of our reality. To what extent this is considered in music therapy has been little studied. Four German and six English language music therapy journals were selected for a literature review of the period 1990–2019. They were systematically screened for articles with an intercultural context. The 196 articles identified, 138 of which in English, were evaluated using time-series representation and an evaluation matrix with nine categories, including client age groups. Intercultural music therapy articles with case studies appear from the turn of the millennium. With an increase in tendency, they have lately accounted for a fluctuation rate of approximately 4 %. The results show a classification of migration-specific pressures in intercultural music therapy work emphasizing the effects on children and adolescents. Pilot studies in multicultural schools and kindergartens report positive effects of music therapy, but admit insufficient sample size and lack of evidence. The importance of intercultural competency for music therapy work is emphasized, however there is no indication of its integration in music therapy training. Interdisciplinary research projects, together with educational and systemic psychiatric disciplines at schools, could help with difficulties in providing evidence for music therapy. Surveys on teaching intercultural competency at music therapy training-institutions would be useful.

A Data Representation Model for Personalized Medicine

Personalized medicine exploits the patient data, for example, genetic compositions, and key biomarkers. During the data mining process, the key challenges are the information loss, the data types heterogeneity and the time series representation. In this paper, a novel data representation model for personalized medicine is proposed in light of these challenges. The proposed model will account for the structured, temporal and non-temporal data and their types, namely, numeric, nominal, date, and Boolean. After the "Date and Boolean" data transformation, the nominal data are treated by dispersion while several clustering techniques are deployed to control the numeric data distribution. Ultimately, the transformation process results in three homogeneous representations with these representations having only two dimensions to ease the exploration of the represented dataset. Compared to the Symbolic Aggregate Approximation technique, the proposed model preserves the time-series information, conserves as much data as possible and offers multiple simple representations to be explored.

Siamese Time Series and Difference Networks for Performance Monitoring in the Froth Flotation Process

Accurate and in-time performance monitoring plays a great role in industrial processes. However, since the labeled performances are usually measured by some special devices with a relatively long-time interval, the current deep neural networks for the performance monitoring always treat them only as an output. But actually, in the industrial process like froth flotation, labeled performance at the previous moment could also offer valuable information for performance monitoring at the current moment, especially under unstable working conditions. Therefore, we propose a Siamese time series and difference network (STS-D net), which integrates input features at different time steps and labeled performance at the previous moment effectively. In this article, the proposed STS-D net includes two sub-networks. One is the Siamese time series network, which aims to extract effective and uniform feature representations for the input time series at the current and previous moments; the other is the difference network, which integrates the feature representations of the two input time series with labeled performance at the previous moment to predict the performance at the current moment in an incremental way. Effectiveness of the proposed STS-D net is validated in a real-world froth flotation process.

Three-Dimensional Structure of the Equatorial Kelvin Wave: Vertical Structure Functions, Equivalent Depths, and Frequency and Wavenumber Spectra

Abstract The wavenumber filtering of the Kelvin waves (KWs) is performed by a multivariate projection of the ERA5 data on the horizontal structures of the KWs from linear theory and vertical structure functions (VSFs) spanning the troposphere and stratosphere. The associated equivalent depths are consistent with solutions expected for the bounded atmosphere. The zonal wavenumber spectra of the KW energy and temporal variance are shown to be continuous and red with the rotational kinetic energy being dominant over the divergent kinetic energy at zonal wavenumber k = 1. Spatial and temporal variance in wavenumber space is analyzed in terms of periods from linear theory. The strongest variability is found in periods of 8–12 days at k = 1 and VSFs with a single zero-crossing in the troposphere and a structure resembling upward propagating waves across the tropopause and higher up. Secondary maxima are associated with VSFs that have multiple zero crossings in the troposphere and small equivalent depths. A comparison of the wavenumber and wavenumber–frequency methods shows a disagreement between frequencies from the Fourier time series and linear theory. A phase-locking experiment, in which the KW phases are replaced by linear theory estimates, demonstrates that power spectra from wavenumber–frequency filtering in the troposphere are largely defined by wave phases.

A spatio-temporal study of state-wide case-fatality risks during the first wave of the COVID-19 pandemic in Mexico

spatio-temporal analysis of the first wave of the coronavirus (COVID-19) pandemic in Mexico (April to September 2020) was performed by state. Descriptive analyses through diagrams, mapping, animations and time series representations were carried out. Greater risks were observed at certain times in specific regions. Various trends and clusters were observed and analysed by fitting linear mixed models and time series clustering. The association of co-morbidities and other variables were studied by fitting a spatial panel data linear model (SPLM). On average, the greatest risks were observed in Baja California Norte, Chiapas and Sonora, while some other densely populated states, e.g., Mexico City, had lower values. The trends varied by state and a four-order polynomial, including fixed and random effects, was necessary to model them. The most common risk development was observed in states belonging to two clusters and consisted of an initial increase followed by a decrease. Some states presented cluster configurations with a retarded risk increase before the decrease, while the risk increased throughout the time of study in others. A cyclic behaviour with a second increasing trend was also observed in some states. The SPLM approach revealed a positive significant association with respect to case fatality risk between certain groups, such as males and individuals aged 50 years and more, and the prevalence of chronic kidney disease, cardiovascular disease, asthma and hypertension. The analysis may provide valuable insight into COVID-19 dynamics applicable in future outbreaks, as well as identify determinants signifying certain trends at the state level. The combination of spatial and temporal information may provide a better understanding of the fatalities due to COVID-19.

Modeling Impulsive Ball Mill Forces Effects on the Dynamic Behavior of a Single-Stage Gearbox

Gearboxes are frequently used in the mining industry, especially for power transmission between the electric drive and the ball mill; besides the extreme complexity of a ball mill gear transmission system, the fault diagnosis by vibration analysis can be easily distorted by the presence of impulsive noises due to the ball pulses on the mill shell. Although several works in the literature are related to the influence of an impulsive noise on the accuracy of the diagnosis, no dynamic model exists yet in the literature that can explain the influence of these forces on the dynamic behavior of gearboxes. This paper presents a new approach to determine the influence of the grinding forces in crack defects diagnosis. This approach is based on a hybrid numerical model of a 24-degree-of-freedom gearbox, simulating one gear train and two drive shafts. The impact forces of the mill drum are modelled by a discrete element method (DEM). The ball-filling rate (Fr), the mill speed (Nr), and the ball size (Db) are considered to study this phenomenon. The simulations results show by a time series representation, fast Fourier transform, and short-time Fourier transform (STFT), that the acceleration is significantly affected by the presence of the grinding forces, developing an impulsive noise due to the impact of the balls governed by the studied parameters.

TimeCLR: A self-supervised contrastive learning framework for univariate time series representation

Time series are usually rarely or sparsely labeled, which limits the performance of deep learning models. Self-supervised representation learning can reduce the reliance of deep learning models on labeled data by extracting structure and feature information from unlabeled data and improve model performance when labeled data is insufficient. Although SimCLR has achieved impressive success in the computer vision field, direct applying SimCLR to time series field usually performs poorly due to the part of data augmentation and the part of feature extractor not being adapted to the temporal dependencies within the time series data. In order to obtain high-quality time series representations, we propose TimeCLR, a framework which is suitable for univariate time series representation, by combining the advantages of DTW and InceptionTime. Inspired by the DTW-based k-nearest neighbor classifier, we first propose the DTW data augmentation that can generate DTW-targeted phase shift and amplitude change phenomena and retain time series structure and feature information. Inspired by the current state-of-the-art deep learning-based time series classification method, InceptionTime, which has good feature extraction capabilities, we designed a feature extractor capable of generating representations in an end-to-end manner. Finally, combining the advantages of DTW data augmentation and InceptionTime, our proposed TimeCLR method successfully extends SimCLR and applies it to the time series field. We designed a variety of experiments and performed careful ablation studies. Experimental results show that our proposed TimeCLR method can not only achieve comparable performance to supervised InceptionTime on multiple tasks, but also produce better performance than supervised learning models in the case of insufficient labeled data, and can be flexibly applied to univariate time series data from different domains.

Development of the GLASS 250-m leaf area index product (version 6) from MODIS data using the bidirectional LSTM deep learning model

Leaf area index (LAI) is a terrestrial essential climate variable that is required in a variety of ecosystem and climate models. The Global LAnd Surface Satellite (GLASS) LAI product has been widely used, but its current version (V5) from Moderate Resolution Imaging Spectroradiometer (MODIS) data has several limitations, such as frequent temporal fluctuation, large data gaps, high dependence on the quality of surface reflectance, and low computational efficiency. To address these issues, this paper presents a deep learning model to generate a new version of the LAI product (V6) at 250-m resolution from MODIS data from 2000 onward. Unlike most existing algorithms that estimate one LAI value at one time for each pixel, this model estimates LAI for 2 years simultaneously. Three widely used LAI products (MODIS C6, GLASS V5, and PROBA-V V1) are used to generate global representative time-series LAI training samples using K-means clustering analysis and least difference criteria. We explore four machine learning models, the general regression neural network (GRNN), long short-term memory (LSTM), gated recurrent unit (GRU), and Bidirectional LSTM (Bi-LSTM), and identify Bi-LSTM as the best model for product generation. This new product is directly validated using 79 high-resolution LAI reference maps from three in situ observation networks. The results show that GLASS V6 LAI achieves higher accuracy, with a root mean square (RMSE) of 0.92 at 250 m and 0.86 at 500 m, while the RMSE is 0.98 for PROBA-V at 300 m, 1.08 for GLASS V5, and 0.95 for MODIS C6 both at 500 m. Spatial and temporal consistency analyses also demonstrate that the GLASS V6 LAI product is more spatiotemporally continuous and has higher quality in terms of presenting more realistic temporal LAI dynamics when the surface reflectance is absent for a long period owing to persistent cloud/aerosol contaminations. The results indicate that the new Bi-LSTM deep learning model runs significantly faster than the GLASS V5 algorithm, avoids the reconstruction of surface reflectance data, and is resistant to the noises (cloud and snow contamination) or missing values contained in surface reflectance than other methods, as the Bi-LSTM can effectively extract information across the entire time series of surface reflectance rather than a single time point. To our knowledge, this is the first global time-series LAI product at the 250-m spatial resolution that is freely available to the public (www.geodata.cn and www.glass.umd.edu).

Fractal Modeling of Polymer Plasma Laser Ablation, Plasma Plume Tsallis Entropy and Its q-Statistics Interpretation, Part I: Theory.

Polymer plasma produced by laser ablation is investigated in a theoretical manner. In relation to the fact that the charge carrier circulation is assumed to take place on fractal curves, the so-called fractality type, electrical charge transport can be resolved by an extended scale relativity method. In addition, an elegant mathematical model, utilizing a conjecture of fractal space-time, is elaborated. The complete solution and its graphical representation for temperature distribution in two-dimensional and three-dimensional cases are successfully introduced. The discrete physical behavior and irrevocable transformation of nanoscale microdomain substructures by laser ablation are realistically examined. Further, benefiting from the interpretation of the fractal analysis, each of the experimental results can be fairly explained. On top of that, this paper presents a proof of Tsallis nonextensive q-statistics, especially for the plasma plume studied. Tsallis entropy in direct connection with fractal dynamics and chaotic-type mechanics of the plasma plume and time-series representation of plasma temperature is introduced for the first time in the present publication, and the q-statistics of the plume plasma temperature are also studied, among others.

Transfer of Statistical Customer Data into Relevant Parameters for the Design of Vehicle Drive Systems

Vehicle drive systems are often oversized for common customer operation in order to cover the high demands of rare driving events such as towing a trailer, high acceleration or steep inclines. This high torque and power requirement affects the efficiency map and the highest efficiency is around the area of increased torque and speed. However, in everyday use, drive systems are mostly driven by customers at low speed and load, and therefore are not operating in the most efficient area. Designing a drive system that only covers the area of highest customer operation can increase efficiency by moving the sweet spot of efficiency to the relevant area, and thus reduce energy consumption. Therefore, customer data need to be analyzed in order to identify customer requirements and to localize the area of greatest operation. The method presented in this paper analyzes customer data in order to identify design-relevant parameters for a customer-specific drive system design. The available customer data results from event-based counts and are submitted as a statistical frequency distribution. These statistics are compared with discrete time series recorded during test drives in order to derive representative time series that correspond to customer behavior. By applying the time frame-based load analysis to these relevant time series, the desired design-relevant parameters are pointed out.

A time series representation of protein sequences for similarity comparison

Based on the physicochemical indexes of 20 amino acids and the Hungarian algorithm, each amino acid was mapped into a vector. And, the protein sequence can be represented as time series in eleven-dimensional space. In addition, the DTW algorithm was applied to calculate the distance between two time series to compare the similarities of protein sequences. The validity and accuracy of this method was illustrated by similarity comparison of ND5 proteins of nine species. Furthermore, homology analysis of eleven ACE2 proteins, which included human, Malayan pangolin and six species of bats, confirmed that the human had shorter evolutionary distance from the pangolin than those bats. The phylogenetic tree of spike protein sequences of 36 coronaviruses, which were divided into five groups, Class I, Class II, Class III, SARS-CoVs and COVID-19, was constructed.

Wind speed forecasting based on model selection, fuzzy cluster, and multi-objective algorithm and wind energy simulation by Betz's theory

Wind energy is of increasing interest to wind farm administrators as a clean and renewable energy source. Accurate wind speed forecasting and effective wind energy simulation can increase the capability of wind power combined with a grid and decrease the operating cost of wind farms. However, many previous studies have been restricted to wind speed forecasting, ignoring wind energy simulations. Thus, grid management cannot effectively estimate the power production of wind farms and leads to an increase in the abandonment wind rate in wind farms. In this study, a wind farm auxiliary management system is developed, which includes two modules: wind speed forecasting and wind energy simulation. In the wind speed forecasting module, first, a data mining algorithm is used to analyze different features of wind speed time series data in a wind farm. Subsequently, a feature selection algorithm is used to determine the representative wind speed time series of the wind farm, and it is combined with a data preprocessing method to effectively eliminate the noise of the original wind speed time series. Second, six hybrid neural network forecasting models based on a modified multi-objective algorithm are established to forecast wind speed. Finally, they are combined with a model selection strategy to yield the best forecasting value for each time point. In the wind energy simulation module, using Betz's theory, the physical transformation process of a wind turbine is estimated to determine the range of wind power generation.

Data Mining and Deep Learning for Predicting the Displacement of "Step-like" Landslides.

Landslide displacement prediction is one of the unsolved challenges in the field of geological hazards, especially in reservoir areas. Affected by rainfall and cyclic fluctuations in reservoir water levels, a large number of landslide disasters have developed in the Three Gorges Reservoir Area. In this article, the Baishuihe landslide was taken as the research object. Firstly, based on time series theory, the landslide displacement was decomposed into three parts (trend term, periodic term, and random term) by Variational Mode Decomposition (VMD). Next, the landslide was divided into three deformation states according to the deformation rate. A data mining algorithm was introduced for selecting the triggering factors of periodic displacement, and the Fruit Fly Optimization Algorithm–Back Propagation Neural Network (FOA-BPNN) was applied to the training and prediction of periodic and random displacements. The results show that the displacement monitoring curve of the Baishuihe landslide has a “step-like” trend. Using VMD to decompose the displacement of a landslide can indicate the triggering factors, which has clear physical significance. In the proposed model, the R2 values between the measured and predicted displacements of ZG118 and XD01 were 0.977 and 0.978 respectively. Compared with previous studies, the prediction model proposed in this article not only ensures the calculation efficiency but also further improves the accuracy of the prediction results, which could provide guidance for the prediction and prevention of geological disasters.

A Novel Time Series Representation Approach for Dimensionality Reduction

With the growth of streaming data from many domains such as transportation, finance, weather, etc, there has been a surge in interest in time series data mining. With this growth and massive amounts of time series data, time series representation has become essential for reducing dimensionality to overcome the available memory constraints. Moreover, time series data mining processes include similarity search and learning of historical data tasks. These tasks require high computation time, which can be reduced by reducing the data dimensionality. This paper proposes a novel time series representation called Adaptive Simulated Annealing Representation (ASAR). ASAR considers the time series representation as an optimization problem with the objective of preserving the time series shape and reducing the dimensionality. ASAR looks for the instances in the raw time series that can represent the local trends and neglect the rest. The Simulated Annealing optimization algorithm is adapted in this paper to fulfill the objective mentioned above. We compare ASAR to three well-known representation approaches from the literature. The experimental results have shown that ASAR achieved the highest reduction in the dimensions. Moreover, it has been shown that using the ASAR representation, the data mining process is accelerated the most. The ASAR has also been tested in terms of preserving the shape and the information of the time series by performing One Nearest Neighbor (1-NN) classification and K-means clustering, which assures its ability to preserve them by outperforming the competing approaches in the K-means task and achieving close accuracy in the 1-NN classification task.