Stationary Time Series Research Articles

Modelability across time as a signature of identity construction on YouTube

Linguistic self-representation and identity construction on social media have attracted much scholarly attention. However, relevant studies tend to overlook the temporal dimension of social media, potentially systematic patterning of linguistic behavior across time, and the attendant implications of such a temporal perspective on identity. Combining an automated lexical tool (LIWC) and the Box–Jenkins method of statistical time series analysis, this paper shows how the ‘modelability’ of linguistic choices across time can be interpreted as signatures of identity construction and complement existing frameworks for identity analysis. Two levels of modelability are discussed—the availability of a well-fitting time series model as evidence of temporal patterning, and specific parameters of that model interpreted in context. These are demonstrated with a case study of the construction of ‘amateur expertise’ over 109 consecutive makeup tutorial videos on the popular YouTube channel ‘Nikkiestutorials’. Results show that the linguistic display of ‘analytical thinking’ reflects a strategy of ‘short term momentum’, the display of ‘clout’ and ‘authenticity’ a strategy of ‘short term restoration’, while the display of ‘emotional tone’ fluctuates randomly across time. The approach is further discussed in terms of its general principles and potential applications in other contexts of identity and related research.

Reproduction in the Archipelago of the Aegean 1920–2016: Long-Term Trends and Recent Upturns

Little is known about the reproductive behavior of the population in insular Greece because of the lack of reliable and unintermittent time series of vital statistics. By applying methods used in historical demography and by indirectly estimating demographic indexes from census returns we try to overcome data scarcity. Substantial differences were found between the islands and mainland Greece. Differences were also found between different groups of islands, with the North Aegean being more conservative as far as the procreative ethics are concerned. Out of wedlock childbearing is less common in North Aegean islands than in Dodecanese or in the Cyclades.

Favorable Commercial and Health Behavior Impacts of a Healthy Vending Policy at an Australian University

BackgroundThe World Health Organization recognizes universities as an important health-promotion setting, including in healthy food provision. Previous research shows that healthy food retail interventions also need to consider commercial sustainability, including financial outcomes, and should take a holistic approach to consumer experience. ObjectiveOur aim was to determine the health behavior and commercial outcomes of a multicomponent traffic light–based healthy vending policy implemented as one part of a holistic university food policy. The hypothesis was that purchases of less healthy “red” beverages would decrease compared with predicted sales, that purchases of healthier “green” and “amber” alternatives would increase, and that there would be no change in revenue. DesignA quasi-experimental design evaluated a real-world food policy using monthly aggregated sales data to compare pre-intervention (January 2016 to March 2018) and post-intervention period sales (December 2018 to December 2019). Participants/settingElectronic sales data were collected from 51 beverage vending machines across 4 university campuses in Victoria, Australia. InterventionA multicomponent policy was implemented between April and November 2018. Beverages were classified using a voluntary state government traffic light framework. Policy included display ≤20% red beverages and ≥50% green beverages; machine traffic light labeling; health-promoting machine branding; review of machine placement; and recycled bottle packaging. Main outcome measuresChanges in red, amber, and green volume sales, and revenue compared with predicated sales. Statistical analyses performedInterrupted time series analysis of sales data compared post-policy sales with predicted sales. ResultsIn the 13th month post-policy implementation, there was a 93.2% (95% CI +35.9% to +150.5%) increase in total beverage volume sold and an 88.6% (95% CI +39.2% to +138.1%) increase in revenue. There was no change in red beverage volume sold, but increases in green (+120.8%; 95% CI +59.0% to +182.6%) and amber (+223.2%; 95% CI +122.4% to +323.9%) volume sold. ConclusionsSustained behavior change and commercial outcomes suggest that holistic vending interventions can effectively promote healthier beverage sales.

Operationalizing a Pandemic-Ready, Telemedicine-Enabled Drive-Through and Walk-In Coronavirus Disease Garage Care System as an Alternative Care Area: A Novel Approach in Pandemic Management

ObjectiveEmergency departments face unforeseen surges in patients classified as low acuity during pandemics such as the coronavirus disease pandemic. Streamlining patient flow using telemedicine in an alternative care area can reduce crowding and promote physical distancing between patients and clinicians, thus limiting personal protective equipment use. This quality improvement project describes critical elements and processes in the operationalization of a telemedicine-enabled drive-through and walk-in garage care system to improve ED throughput and conserve personal protective equipment during 3 coronavirus disease surges in 2020. MethodsStandardized workflows were established for the operationalization of the telemedicine-enabled drive-through and walk-in garage care system for patients presenting with respiratory illness as quality improvement during disaster. Statistical control charts present interrupted time series data on the ED length of stay and personal protective equipment use in the week before and after deployment in March, July, and November 2020. ResultsPhysical space, technology infrastructure, equipment, and staff workflows were critical to the operationalization of the telemedicine-enabled drive-through and walk-in garage care system. On average, the ED length of stay decreased 17%, from 4.24 hours during the week before opening to 3.54 hours during the telemedicine-enabled drive-through and walk-in garage care system operation. There was an estimated 25% to 41% reduction in personal protective equipment use during this time. ConclusionLessons learned from this telemedicine-enabled alternative care area implementation can be used for disaster preparedness and management in the ED setting to reduce crowding, improve throughput, and conserve personal protective equipment during a pandemic.

An Exploratory Study of Geospace Perturbations Using Financial Analysis Tools in the Context of Complex Systems

In the frame of complex systems research, methods used to quantitatively analyze specific dynamic phenomena are often used to analyze phenomena from other disciplines on the grounds that are governed by similar dynamics. Technical analysis is considered the oldest, currently omnipresent, method for financial markets analysis, which uses past prices aiming at the possible short-term forecast of future prices. This work is the first attempt to explore the applicability of technical analysis tools on disturbance storm time (Dst) index time series, aiming at the identification of similar features between the Dst time series during magnetic storms (MSs) and asset price time series. We employ the following financial analysis tools: simple moving average (SMA), Bollinger bands, and relative strength index (RSI), formulating an analysis approach based on various features, appearing in financial time series during high volatility periods, that could be found during the different phases of the evolution of an MS (onset, main development, and recovery phase), focusing on the temporal sequence they occur. The applicability of the proposed analysis approach is examined on several MS events and the results reveal similar behavior with the financial time series in high volatility periods. We postulate that these specialized data analysis methods could be combined in the future with other statistical and complex systems time series analysis methods in order to form a useful toolbox for the study of geospace perturbations related to natural hazards.

Occupational Disease as the Bane of Workers’ Lives: A Chronological Review of the Literature and Study of Its Development in Slovakia. Part 1

This article not only offers a chronological overview of the development of occupational medicine, but also offers a summary of occupational diseases recommended by the ILO and legislative decisions that have influenced how we approach assessment today. We consider that these areas form a whole in which they cannot function without each other and they would lose their relevance if the system was collapsed. By excluding even one part of it, we would find ourselves at the beginning of the era of occupational medicine, and a large number of employees would once again be exposed to conditions that previously led to considerable illness and mortality of employees. The article also examines legislation and the development of occupational diseases in Slovakia in the period 1997–2019. Using basic statistical methods and time series, a trend model for the time series of the development of the number of occupational diseases over the last 20 years is created. The modeling also includes a forecast for the development of the number of occupational diseases for the next 5 years. The model created shows a favorable, decreasing trend in the number of occupational diseases in Slovakia.

A new tendency correlation coefficient for bivariate time series

The correlation analysis of bivariate time series is a common problem in many fields. Based on the analysis of the changing range and the changing direction between any two time points on the bivariate time series, this paper proposes a new tendency correlation coefficient that is applicable for both stationary and non-stationary time series. It ranges from − 1 to 1, and the necessary and sufficient condition of the tendency correlation coefficient getting the extremum value is theoretically given. To measure the similarity of bivariate time series, the concept of congruence is proposed based on the normalization process, and the invariant character of the tendency correlation coefficient for the congruent bivariate time series is theoretically proved. Compared with other existing methods, experiment results of Monte Carlo simulations on both stationary series and non-stationary series with different degrees of intrinsic correlations, and real time series of oil-immersed transformers convince the advantages of the proposed tendency correlation coefficient on accuracy and stability.

Stuck in the past or living in the present? Temporal focus and the spread of COVID-19

Research has shown that the temporal focus of individuals can have a real effect on behavior. In the context of the COVID-19 pandemic, this study posits that temporal focus will affect adherence behavior regarding health control measures, such as social distancing, hand washing and mask wearing, which will be manifested through the degree of spread of COVID-19. It is suggested that social media can provide an indicator of the general temporal focus of the population at a particular time. In this study, we examine the temporal focus of Twitter text data and the number of COVID-19 cases in the US over a 317-day period from the inception of the pandemic, using text analytics to classify the temporal content of 0.76 million tweets. The data is then analyzed using dynamic regression via advanced ARIMA modelling, differencing the data, removing weekly seasonality and creating a stationary time series. The result of the dynamic regression finds that past orientation does indeed have an effect on the growth of COVID-19 cases in the US. However, a present focus tends to reduce the spread of COVID cases. Future focus had no effect in the model. Overall, the research suggests that detecting and managing temporal focus could be an important tool in managing public health during a pandemic.

A new framework for a multi-site stochastic daily rainfall model: Coupling a univariate Markov chain model with a multi-site rainfall event model

Multi-site rainfall models are useful tools to provide synthetic realizations of spatially-correlated rainfall at multiple stations, which are of great importance for flood and drought risk assessment and climate change impact analysis. Therefore, a good preservation of various observed rainfall characteristics including rainfall time-series statistics and rainfall event characteristics at individual stations and the inter-site correlations of these rainfall characteristics is very crucial. To achieve this purpose, this study aims to develop a multi-site stochastic daily rainfall model by coupling a univariate Markov chain with a multi-site rainfall event model (MSDRM-MCREM), based on our previously-developed single-site SDRM-MCREM. The univariate Markov chain model in MSDRM-MCREM is used to generate spatially-correlated multi-site rainfall occurrence time series and extract simulated rainfall events for individual stations based on continuous wet days. The multi-site rainfall event model is then constructed using Vine copulas to simulate spatially-correlated rainfall event characteristics of those simulated rainfall events that occur simultaneously at multiple stations, including rainfall durations, rainfall depths and temporal patterns. Subsequently, this model was applied to the Changshangang River basin in Zhejiang Province, East China and its performance in reproducing rainfall characteristics and spatial correlations was evaluated for three cases, i.e. simulations for two, three and four stations. Results show that except for overestimation of light rainfall, MSDRM-MCREM can simultaneously well preserve rainfall time-series statistics (i.e. different rainfall percentiles, mean monthly rainfall, standard deviations and probabilities and mean values of wet days), extreme rainfall (i.e. exceedance probabilities of annual maximum 1-day, 3-day and 5-day rainfall) and rainfall event characteristics (i.e. cumulative probabilities of wet spell, dry spell and rainfall depth, temporal patterns and occurrence probabilities of rainfall types for different depth-based event classes) at individual stations. In addition, the spatial correlations of rainfall characteristics have also been well maintained, including rainfall occurrence time series and rainfall event characteristics in different groups, with the inter-site correlations of rainfall time series being slightly underestimated.

SPHARMA approximations for stationary functional time series on the sphere

In this paper, we focus on isotropic and stationary sphere-cross-time random fields. We first introduce the class of spherical functional autoregressive-moving average processes (SPHARMA), which extend in a natural way the spherical functional autoregressions (SPHAR) recently studied in Caponera and Marinucci (Ann Stat 49(1):346–369, 2021) and Caponera et al. (Stoch Process Appl 137:167–199, 2021); more importantly, we then show that SPHAR and SPHARMA processes of sufficiently large order can be exploited to approximate every isotropic and stationary sphere-cross-time random field, thus generalizing to this infinite-dimensional framework some classical results on real-valued stationary processes. Further characterizations in terms of functional spectral representation theorems and Wold-like decompositions are also established.

A note on time series differencing

Differencing is one of the key tools time series analysis. It is commonly used to obtain stationary time series. In this note, we show that the nth difference of a weakly stationary time series is weakly stationary. Similarly we prove that the nth difference of a strictly stationary time series is strictly stationary. We also consider the effect of differencing on the time series autocovariance.

Assessment of properties of vegetation phenology in fire-affected areas from 2000 to 2015in the Peloponnese, Greece

Fire events are among the natural disasters that affect both ecosystem functioning but also human lives. The present work deals with the assessment of the properties of vegetation phenology in areas affected by fire events during a 10-year period in the Peloponnese, Greece. The main objective is to identify special patterns in NDVI time series of fire-affected areas, that can be used either as precursory indicators or as a tool for assessment of regeneration patterns. Therefore, we constructed and analyzed time series of remotely sensed Moderate Resolution Imaging Spectroradiometer (MODIS) NDVI at 250 m spatial resolution with a 16-day time step for affected and non-affected areas surrounding the burnt sites, where the majority of the area refers to the same land cover type. Analysis of changes of NDVI time series of both burnt and unburnt areas was conducted and results indicated that all fire events in the burnt sites caused breaks in the NDVI time series, demonstrated either as NDVI shifts to the lower and/or as changes in the trend and seasonality of the time series. Trends of the NDVI time series of the affected sites were almost doubled in the post fire period compared to the period before the event, indicating the evolution of the regenerating process. The basic statistics of the NDVI time series of the affected and non-affected areas correspond to greater minimum NDVI values of the affected areas throughout the study period, except of the two years after the fire event. Mean NDVI values were almost equal in burnt and unburnt areas before the events and reached the same values approximately seven years after the fire. Maximum NDVI values of the affected and non-affected sites were almost equal before the event but were abruptly reduced in the affected areas after the events and revert to their original state approximately ten years after. An alarming finding is the considerably lower standard deviation of the fire affected areas in comparison to non-fire-affected ones, during all years of the study period, except of a short period immediately after the fire event. It seems thus that areas covered by homogeneous vegetation with no fire breaks or other landforms to act as barriers to fire are the most vulnerable and most severely affected.

Data complexity based hybrid approach to select a model for wind speed forecasting

ABSTRACT Recent wind speed forecasting works have applied various time series decomposition methods to segregate a complex wind time series into linear and nonlinear components. Then, existing statistical techniques are used to model each component separately. This approach is reliable and produces accurate results when applied to complex wind time series. However, forecasting the stationary wind time series may produce unsatisfactory results. In order to address this problem, we propose a hybrid approach for wind speed forecasting. In this approach, seven different data complexity measures are used to calculate time series data complexity. Then, the measured complexity is used by the proposed model to decide whether to go with the decomposition technique or not. This paper has suggested Empirical Mode Decomposition for decomposition, which has achieved an incredible surge in a large number of data streams. Similarly, two leading statistical forecasting approaches, i.e., Autoregressive Integrated Moving Average and Generalized Autoregressive Score, are employed to model linear time series and decomposed components of the complex nonlinear time series. For comparative analysis, Mean Absolute Error, Root Mean Squared Error, and R-square are used as evaluation metrics. Experimental result in this work suggests that the proposed hybrid approach performs equivalently well for complex time series and achieves considerable improvement for stationary wind time series. The percentage improvement is 95.29% in the case of ARIMA and 49.09% when GAS is applied.

Fast parallel Newton–Raphson power flow solver for large number of system calculations with CPU and GPU

To analyze large sets of grid states, e.g. when evaluating the impact from the uncertainties of the renewable generation with probabilistic Monte Carlo simulation or in stationary time series simulation, large number of power flow calculations have to be performed. For the application in real-time grid operation, grid planning and in further cases when computational time is critical, a novel approach on simultaneous parallelization of many Newton–Raphson power flow calculations on CPU and with GPU-acceleration is proposed. The result shows a speed-up of over x100 comparing to the open-source tool pandapower, when performing repetitive power flows of system with admittance matrix of the same sparsity pattern on both CPU and GPU. The speed-up relies on the algorithm improvement and highly optimized parallelization strategy, which can reduce the repetitive work and saturate the high hardware computational capability of modern CPUs and GPUs well. This is achieved with the proposed batched sparse matrix operation and batched linear solver based on LU-refactorization. The batched linear solver shows a large performance improvement comparing to the state-of-the-art linear system solver KLU library and a better saturation of the GPU performance with small problem scale. Finally, the method of integrating the proposed solver into pandapower is presented, thus the parallel power flow solver with outstanding performance can be easily applied in challenging real-life grid operation and innovative researches e.g. data-driven machine learning studies.

The identification of artefacts in reporting of drug‐induced deaths using structural breaks analysis of time series statistics

Drug-related mortality is a key epidemiological indicator that is collected nationally and internationally. Significant efforts were made in 2006-2007 to improve the quality of data concerning drug-related mortality in the Czech Republic. The aim of this article is to identify the effect of a quality improvement project on the drug-induced mortality data reported in the General Mortality Registry (GMR), and to demonstrate how to identify, quantify and interpret changes in drug-induced mortality based on the example of the Czech Republic. We extracted data on illicit drug-induced deaths from the Czech Republic GMR and Special Mortality registry (SMR) for the years between 2004 and 2012, and aggregated monthly and quarterly time series. We applied a new procedure to identify structural breakpoints in time series based on dating structural changes in standard linear regression models. In the GMR, breakpoints were identified in three time series: (i) opioid-related deaths; (ii) other stimulant-related deaths; and (iii) total drug-induced deaths. In the SMR, the structural breaks were identified for opioids, volatile substances and selection D time series. In each of these time series, the analysis identified a decrease in the intercepts in the different segments. The structural breaks identified and quantified in the GMR time series were plausibly caused by the quality improvement efforts that started in 2006. These results demonstrate that it is critical for the analysis and use of drug mortality data collected in the registries to identify practice changes in the relevant registries, to quantify their influence and to adjust mortality estimates accordingly.

The four models for forecasting the peak period of Dendrolimus punctatus (Lepidoptera: Lasiocampiade) for the second generation egg

AbstractTo improve the accuracy of forecasting the peak period of Dendrolimus punctatus, stationary time series, periodic distance method, stepwise regression model and the Bayes discriminant analysis were used. RSME value, kappa coefficient and accuracy were used as evaluation criteria to predict the peak period for the second generation egg of D. punctatus with over 33 years from 1983 to 2016 in Qianshan County, Anhui Province. The predictions of these models were verified in 2017 and 2018. The prediction of the stationary time model and Bayes discriminant analysis for 2017 was one level lower than the actual result and for 2018 was one level higher than the actual result, while the prediction of the periodic distance method was identical to the actual result for 2017 and greatly different from the actual result for 2018. The accuracy for stationary time series (RMSE = 0.92 kappa = 0.76) and periodic distance method (RMSE = 2.96, kappa = 0.81) from 1983 to 2018 were 87.88% and 85.71%, respectively. When taking into consideration the standard error was based on differential, the accuracy for the prediction of stepwise regression model (RMSE = 0.25, kappa = 1.00) from 1983 to 2018 was 100%. The accuracy of Bayes discriminant method (RMSE = 0.71, kappa = 0.96) was 97.14%. Comparatively speaking, the stepwise regression model and Bayes discriminant analysis method were better than the stationary time series and periodic distance method in RMSE value, kappa coefficient and accuracy. So they were relatively ideal forecast methods.

Understanding and mitigating the impact of data gaps on offshore wind resource estimates

Abstract. Like almost all measurement datasets, wind energy siting data are subject to data gaps that can for instance originate from a failure of the measurement devices or data loggers. This is in particular true for offshore wind energy sites where the harsh climate can restrict the accessibility of the measurement platform, which can also lead to much longer gaps than onshore. In this study, we investigate the impact of data gaps, in terms of a bias in the estimation of siting parameters and its mitigation by correlation and filling with mesoscale model data. Investigations are performed for three offshore sites in Europe, considering 2 years of parallel measurement data at the sites, and based on typical wind energy siting statistics. We find a mitigation of the data gaps' impact, i.e. a reduction of the observed biases, by a factor of 10 on mean wind speed, direction and Weibull scale parameter and a factor of 3 on Weibull shape parameter. With increasing gap length, the gaps' impact increases linearly for the overall measurement period while this behaviour is more complex when investigated in terms of seasons. This considerable reduction of the impact of the gaps found for the statistics of the measurement time series almost vanishes when considering long-term corrected data, for which we refer to 30 years of reanalysis data.

Effect of environmental factors and density-dependence on somatic growth of Eastern Georges Bank haddock (Melanogrammus aeglefinus)

Haddock (Melanogrammus aeglefinus) is an important commercial fish species in the Northwest Atlantic, with the Eastern Georges Bank population a transboundary resources utilized by both Canada and the United States. With the dramatic increase recruitment of this population in the last decade, in particular concentrations of fish on the eastern portion of the Bank, there has been a precipitous decrease in somatic growth, raising concerns for managers and stakeholders. In this study, the effect of environmental factors and density-dependence on somatic growth of eastern Georges Bank haddock was studied using a 2-step Generalized Additive Model (GAM) regression approach. The biotic and abiotic factors examined in the regression analysis included haddock density, sampling location, local sea surface and bottom temperature, salinity and short time series of phytoplankton bloom statistics. The relative importance of each covariate in the best selected model was investigated using the nested GAMs. We found that density-dependence explained most of the variation in haddock growth, where the impact of temperature varied by season. Though of less model importance, annual winter salinity and phytoplankton bloom magnitude (lagged by one year) also contributed the growth model fits. These relationships have the immediate application of contextualizing the observed change in haddock growth, but may also inform various aspects of the stock assessment process.

Evidence for non-Rayleigh characteristics in ship underwater acoustic signatures.

This work examines ship underwater acoustic signature amplitude statistics and statistical distributions and explores the hypothesis that ship signatures exhibit amplitude fluctuations that are different from Rayleigh-distributed ambient ocean noise. Signature measurements from a small oceanographic ship conducting a variety of maneuvers are examined, focusing on those conditions where propeller cavitation and broadband signal modulation occur, specifically during maximum speed runs and turning maneuvers. A key difference for a ship signature is the amplitude modulation generated by propeller cavitation, and this is found to be associated with super-Rayleigh signal characteristics. Cyclostationary processing techniques are used to estimate propeller shaft- and blade-rate modulation. Under conditions of clearly observable propeller modulation, time-series statistics scintillation index and skewness showed values significantly in excess of Rayleigh-distributed values. Ship signature amplitude probability density functions were found to be better matched by a K-distribution model with small shape factor, indicating clearly non-Rayleigh behavior. Evidence for a secondary, high-amplitude, lognormally distributed sound generation mechanism was observed during maximum helm turns.

A Meta-Learning approach to Model Uncertainty in Financial Time Series

Financial markets have experienced several negative sigma events in recent years; these events occur with much more regularity than current risk models can predict. There is no guarantee that the training set's data generating process will be the same in the test set in finance. Mathematical models are designed to operate with unlimited and changing data, and yet, actual events keep making life hard for most models. The assumption of independent and identically distributed random variables and a stationary time series do not hold in reality. Over-reliance on historical data and backtesting of models is not a sufficient approach to overcome these challenges. Reinforcement-learning faces similar challenges when applied to financial time series. Out-of-distribution generalization is a problem that cannot be solved without assumptions on the data generating process. If the test data is arbitrary or unrelated to the training data, then generalization is not possible. Finding these particular principles could potentially help us build AI and financial modeling systems. N-Beats, Oreshkin et al. [2020], is a deep neural architecture with backward and forward residual links and a deep stack of fully-connected layers. N-Beats can be considered as a meta-learning model for time series prediction. Meta-Learning is a machine learning approach that intends to design models that can learn new skills or adapt to new environments rapidly with few training examples. We explore the performance of N-Beats and compare its performance with other deep learning models. The results are not conclusive in establishing N-Beats as a better model than the other models tested in this study. We show in this study that other neural network-based models offer similar performance.