Causal Time Series Research Articles

Neural Dynamics Associated with Biological Variation in Normal Human Brain Regions.

The processes involved in encoding and decoding signals in the human brain are a continually studied topic, as neuronal information flow involves complex nonlinear dynamics. This study examines awake human intracranial electroencephalography (iEEG) data from normal brain regions to explore how biological sex influences these dynamics. The iEEG data were analyzed using permutation entropy and statistical complexity in the time domain and power spectrum calculations in the frequency domain. The Bandt and Pompe method was used to assess time series causality by associating probability distributions based on ordinal patterns with the signals. Due to the invasive nature of data acquisition, the study encountered limitations such as small sample sizes and potential sources of error. Nevertheless, the high spatial resolution of iEEG allows detailed analysis and comparison of specific brain regions. The results reveal differences between sexes in brain regions, observed through power spectrum, entropy, and complexity analyses. Significant differences were found in the left supramarginal gyrus, posterior cingulate, supplementary motor cortex, middle temporal gyrus, and right superior temporal gyrus. This study emphasizes the importance of considering sex as a biological variable in brain dynamics research, which is essential for improving the diagnosis and treatment of neurological and psychiatric disorders.

The link between transportation industry and transportation cryptocurrencies – A comparison among airline, rail, and shipping industries

This work examines the link between some of the most important transportation companies’ stock performance with the corresponding transportation cryptocurrencies. To do so, airlines, rail, and shipping companies’ stocks are investigated, examining for a probable link with the transportation cryptocurrencies, and also investigating the attributes of this link. The variable-Lag time-series causality is employed to test the channel of causality (if such exists), and the multifractal detrended cross-correlation analysis is utilized to investigate the relationship’s attributes. According to the results, the three transportation industries differ in the way they relate to transportation-related cryptocurrencies. More precisely, the transportation industries differ in the way they relate to transportation cryptocurrencies since airlines, and railway companies affect the price of the cryptocurrencies related to transportation, and cryptocurrencies also affect in a similar magnitude these fields. On the other hand, the shipping companies have a greater effect on the cryptocurrencies, while the cryptocurrencies in very few cases affect these stocks. Finally, a long-run relationship is identified, implying that transportation companies and the corresponding cryptocurrencies are positively and long-term related. The results important because they unveil a link between the transportation companies and the corresponding cryptocurrencies, with this link differing among these industries. This finding should be taken into account in the adoption of relevant technologies from the transportation field.

The Role of Currency Fluctuations in Shaping Kenya’s Export Performance

This study explores the impact of exchange rates on Kenya’s export performance from 1990 to 2022. The study utilizes a causal research design and time series data to assess the influence of exchange rate fluctuations, volatility, and relative price levels on export volumes and their contribution to GDP. Descriptive statistics reveal considerable variability, with exports displaying slight positive skewness and exchange rates exhibiting a slightly left-skewed distribution. Unit root tests confirm non-stationarity in the level forms of exports, exchange rates, the export unit value index, and the price level ratio, which become stationary after first differencing, thus suitable for econometric modeling. The Vector Error Correction Model results indicate insignificant long-run effects of exchange rates, export unit value index, andprice level ratio on exports, despite significant short-run adjustments correcting deviations from the long-run equilibrium by approximately 73.4% per period. Short-run analyses show insignificant effects of lagged exchange rate andexport unit value index differences on exports, while price level ratio exhibits a marginally significant negative effect. The research emphasizes the necessity for stable exchange rates to foster a predictable export environment, crucial for attracting investments and sustaining economic growth. The study recommends diversification of Kenya’s export base, enhanced competitiveness through technological advancements and skill development, and creation of financial instruments to mitigate exchange rate risks. Strengthening export promotion initiatives and improving data collection and research are also advocated to support informed policymaking. The study’s findings are pertinent to policymakers, investors, and researchers, highlighting how currency fluctuations affect economic stability, job creation, and living standards.

Generative Representation Learning in Recurrent Neural Networks for Causal Timeseries Forecasting

Generative Representation Learning in Recurrent Neural Networks for Causal Timeseries Forecasting

Seismic Velocity Inversion via Physical Embedding Recurrent Neural Networks (RNN)

Seismic velocity inversion is one of the most critical issues in the field of seismic exploration and has long been the focus of numerous experts and scholars. In recent years, the advancement of machine learning technologies has infused new vitality into the research of seismic velocity inversion and yielded a wealth of research outcomes. Typically, seismic velocity inversion based on machine learning lacks control over physical processes and interpretability. Starting from wave theory and the physical processes of seismic data acquisition, this paper proposes a method for seismic velocity model inversion based on Physical Embedding Recurrent Neural Networks. Firstly, the wave equation is a mathematical representation of the physical process of acoustic waves propagating through a medium, and the finite difference method is an effective approach to solving the wave equation. With this in mind, we introduce the architecture of recurrent neural networks to describe the finite difference solution of the wave equation, realizing the embedding of physical processes into machine learning. Secondly, in seismic data acquisition, the propagation of acoustic waves from multiple sources through the medium represents a high-dimensional causal time series (wavefield snapshots), where the influential variable is the velocity model, and the received signals are the observations of the wavefield. This forms a forward modeling process as the forward simulation of the wavefield equation, and the use of error back-propagation between observations and calculations as the velocity inversion process. Through time-lapse inversion and by incorporating the causal information of wavefield propagation, the non-uniqueness issue in velocity inversion is mitigated. Through mathematical derivations and theoretical model analyses, the effectiveness and rationality of the method are demonstrated. In conjunction with simulation results for complex models, the method proposed in this paper can achieve velocity inversion in complex geological structures.

A novel adaptive multi-scale Rényi transfer entropy based on kernel density estimation

The utilization of Rényi Transfer Entropy (RTE) as a powerful model for analyzing causal relationships among variables has become pervasive in the field of complex systems time series causality detection. However, there are two major challenges in using RTE: the inherent multilevel structure of causal associations in the systems, and the precision in RTE estimation. To address these challenges, this paper proposes an adaptive multi-scale Rényi transfer entropy based on kernel density estimation. The framework of causal detection based on the novel RTE consists of two parts: adaptive discrete wavelet transform (ADWT)-based time series decomposition and multivariate kernel density estimation (MKDE)-based causality network generation. In the ADWT-based time series decomposition, the original series are decomposed into different frequency bands by optimal wavelet coefficients, which is generated adaptively by an auto-encoder. In the MKDE-based causality network generation, the causal network between the variables is represented by an adjacency matrix composed of their decomposition components in each layer, and the values of the matrix are the RTE values between the variables. In order to accurate estimation of RTE values an evaluation criterion for KDE under a uniform measure in both univariate and multivariate cases and the optimal bandwidth selection is provided in this part. To validate the effectiveness of the novel causal measure in this paper, the proposed method is tested on the synthetic and real data, and the results show that it can effectively detect causal relationships among variables at different levels in non-stationary time series of both bidirectional and undirectional complex systems. Compared to the other RTE estimators, the proposed method can detect the causality accurately and avoid the spurious causality.

Time series causal relationships discovery through feature importance and ensemble models

Inferring causal relationships from observational data is a key challenge in understanding the interpretability of Machine Learning models. Given the ever-increasing amount of observational data available in many areas, Machine Learning algorithms used for forecasting have become more complex, leading to a less understandable path of how a decision is made by the model. To address this issue, we propose leveraging ensemble models, e.g., Random Forest, to assess which input features the trained model prioritizes when making a forecast and, in this way, establish causal relationships between the variables. The advantage of these algorithms lies in their ability to provide feature importance, which allows us to build the causal network. We present our methodology to estimate causality in time series from oil field production. As it is difficult to extract causal relations from a real field, we also included a synthetic oil production dataset and a weather dataset, which is also synthetic, to provide the ground truth. We aim to perform causal discovery, i.e., establish the existing connections between the variables in each dataset. Through an iterative process of improving the forecasting of a target’s value, we evaluate whether the forecasting improves by adding information from a new potential driver; if so, we state that the driver causally affects the target. On the oil field-related datasets, our causal analysis results agree with the interwell connections already confirmed by tracer information; whenever the tracer data are available, we used it as our ground truth. This consistency between both estimated and confirmed connections provides us the confidence about the effectiveness of our proposed methodology. To our knowledge, this is the first time causal analysis using solely production data is employed to discover interwell connections in an oil field dataset.

Effect of Recent Abortion Legislation on Twitter User Engagement, Sentiment, and Expressions of Trust in Clinicians and Privacy of Health Information: Content Analysis.

The Supreme Court ruling in Dobbs v Jackson Women's Health Organization (Dobbs) overrules precedents established by Roe v Wade and Planned Parenthood v Casey and allows states to individually regulate access to abortion care services. While many states have passed laws to protect access to abortion services since the ruling, the ruling has also triggered the enforcement of existing laws and the creation of new ones that ban or restrict abortion. In addition to denying patients the full spectrum of reproductive health care, one major concern in the medical community is how the ruling will undermine trust in the patient-clinician relationship by influencing perceptions of the privacy of patient health information. This study aimed to study the effect of recent abortion legislation on Twitter user engagement, sentiment, expressions of trust in clinicians, and privacy of health information. We scraped tweets containing keywords of interest between January 1, 2020, and October 17, 2022, to capture tweets posted before and after the leak of the Supreme Court decision. We then trained a Latent Dirichlet Allocation model to select tweets pertinent to the topic of interest and performed a sentiment analysis using Robustly Optimized Bidirectional Encoder Representations from Transformers Pre-training Approach model and a causal impact time series analysis to examine engagement and sentiment. In addition, we used a Word2Vec model to study the terms of interest against a latent trust dimension to capture how expressions of trust for our terms of interest changed over time and used term frequency, inverse-document frequency to measure the volume of tweets before and after the decision with respect to the negative and positive sentiments that map to our terms of interest. Our study revealed (1) a transient increase in the number of daily users by 576.86% (95% CI 545.34%-607.92%; P<.001), tweeting about abortion, health care, and privacy of health information postdecision leak; (2) a sustained and statistically significant decrease in the average daily sentiment on these topics by 19.81% (95% CI -22.98% to -16.59%; P=.001) postdecision leak; (3) a decrease in the association of the latent dimension of trust across most clinician-related and health information-related terms of interest; (4) an increased frequency of tweets with these clinician-related and health information-related terms and concomitant negative sentiment in the postdecision leak period. The study suggests that the Dobbs ruling has consequences for health systems and reproductive health care that extend beyond denying patients access to the full spectrum of reproductive health services. The finding of a decrease in the expression of trust in clinicians and health information-related terms provides evidence to support advocacy and initiatives that proactively address concerns of trust in health systems and services.

Russian Influence Operations during the Invasion of Ukraine

Prior to their invasion of Ukraine, the Russian Federation was seen as having a vastly superior ability to conduct operations in the information environment, particularly their ability to influence foreign audiences, when compared to their western counterparts (Cunningham, 2020). However, the efforts of Ukraine, NATO allies, and other aligned nations to conduct operations in the information environment with intelligence pre-bunking, traditional diplomacy, and sanctions, blunted many of Russia’s best efforts. Further tactical and operational failures from the Armed Forces of the Russian Federation also undercut the salience of Russian messaging campaigns. This does not mean that Russia’s efforts in the informational environment fully failed or did not adapt to these actions in the lead-up and continuation of the war. Through the use of the Natural Language Processing (NLP) technique transformer-based topic modeling (Grootendorst, 2022) and the causal inference technique Bayesian Structural Time Series analysis (Brodersen, 2015) this paper looks to both qualitatively and quantitatively examine how Russian state media on Twitter reacted, changed narratives, and focused efforts regionally from January 1st through September 1st, 2022. Through this analysis we argue that Russian efforts in Europe may have been of limited success. We further argue that by shifting focus, Russia gained influence in South America, and Middle East and North Africa, where their influence operations faced minimal obstructions, such as sanctions, and a latent anti-western sentiment.

Nonlinear time-series analysis on the causality response of phytoplankton in a tributary of the Three Gorges Reservoir

水华的频发已成为当前三峡水库最为突出的生态环境问题之一。尽管水动力调控叠加上温度变暖和营养负荷增加会诱导水华暴发强度和频率增加,但仍缺乏有效的方法框架去利用野外观测数据评估环境因子与浮游植物间的因果关联。本研究以三峡水库澎溪河监测数据为例,采用非线性时序分析的建模框架来量化浮游植物的因果响应规律。数据来自于2007年6月至2018年9月澎溪河流域的高阳平湖和汉丰湖两个观测点,其中包含了水文、气象和水质及叶绿素a等11种变量。首先,利用奇异谱分析(SSA)分离了叶绿素a和环境因子的低维确定性动力学信号;其次,采用收敛交叉映射(CCM)方法检验了叶绿素a与环境因子间的因果关联。结果显示:① 气象因子、支流流量、水温、三峡大坝水位和上游调节坝水位是影响高阳平湖叶绿素a时序变化的重要因素;② 总氮、总磷为代表的营养盐只在汉丰湖观测点中表现出与叶绿素a的因果关系,且总氮较之于总磷对叶绿素a变化影响更为显著;③ CCM结果与传统的皮尔森相关性分析及格兰杰因果检验比较,证实非线性时序分析方法在分析浮游植物的因果响应上更具优势。本研究为水生态系统的因果建模提供了研究范例,也为推动利用长期观测数据评估三峡水库水华驱动因子提供了新的研究角度。;Harmful algae blooms in Three Gorges Reservoir (TGR) have become the most major environmental and ecological problem. Despite broad recognition that hydrological manipulation, warming climate and nutrient loading promote the intensity and frequency of bloom events, the main problem is still no science-based framework for evaluating the causal relationships between environmental factors and phytoplankton biomass by using field observation data. Thus, as a case study, a nonlinear time-series causality analysis framework was used to reveal the phytoplankton response in the TGR, China. Two sampling stations, named Lake Gaoyangping and Lake Hanfeng, were used to reconstruct the dynamics of real-world environmental systems with eleven parameters from June 2007 to September 2018. Singular spectrum analysis isolated low-dimensional, deterministic signals for chlorophyll-a(Chl.a) dynamics and other candidate drivers. Causal analysis with convergent cross-mapping (CCM) provided strong evidence that temperature, sunshine hours, precipitation, discharge in the tributary, and water level of TGR systematically influenced phytoplankton biomass dynamics in Lake Gaoyangping. In contrast, total nitrogen (TN) and total phosphorus (TP) showed a causal relationship with Chl.a only in Lake Hanfeng, and the effect of TN on Chl.a was more significant than that of TP. Finally, the research compares the nonlinear CCM analysis with the linear Pearson correlation analysis and Granger causality test. It confirms that the nonlinear causality method has more advantage in causal discovery based on long time-series monitoring data. Overall, this study provides a case for causal modelling of water ecosystems and a new research perspective for evaluating the environmental drivers of harmful algae blooms in TGR by using long-term observational data.

Root cause diagnosis for process faults based on multisensor time-series causality discovery

Root cause diagnosis (RCD) focuses on locating the critical root causes and identifying the propagation paths of the industrial process faults, which is superior to conventional fault diagnosis methods and has attracted extensive attention. However, RCD is challenging in terms of dense sensor layout and diversified interconnected subsystems in complex industrial processes, since a fault originating in one unit can propagate throughout the entire plant along with the flow of information and material, obscuring its root causes. Moreover, the previous RCD methods ignore the more crucial temporal information between multisensor time-series, suffering from the disadvantages of requiring expert knowledge and responding untimely. Therefore, a novel data-driven RCD methodology named multisensor time-series causality discovery (MTCD) is proposed for industrial processes fault diagnosis. Firstly, a temporal registration network (TRN) based on the dilated convolutional neural network (DCNN) is proposed to extract the temporal correlation between time-series and implement the time-series prediction. Ulteriorly, a permutation importance causality validation (PICV) is designed to verify the causality between time-series based on the TRN prediction results without prior knowledge or expert experience. Besides, a causal time delay discovery approach based on layerwise relevance propagation (LRP) is presented by interpreting the weights of the trained TRN. Finally, we demonstrate the effectiveness of the proposed method with a stochastic synthesis process and the Tennessee Eastman process. The experimental results show that the proposed method has better performance in both root cause localization and propagation path identification.

Fault Early Warning Model for High-Speed Railway Train Based on Feature Contribution and Causal Inference

The demands for model accuracy and computing efficiency in fault warning scenarios are increasing as high-speed railway train technology continues to advance. The black box model is difficult to interpret, making it impossible for this technology to be widely adopted in the railway industry, which has strict safety regulations. This paper proposes a fault early warning machine learning model based on feature contribution and causal inference. First, the contributions of the features are calculated through the Shapley additive explanations model. Then, causal relationships are discovered through causal inference models. Finally, data from causal and high-contribution time series are applied to the model. Ablation tests are conducted with the Naïve Bayes, Gradient Boosting Decision Tree, eXtreme Gradient Boosting, and other models in order to confirm the efficiency of the method based on early warning data regarding the on-site high-speed train traction equipment circuit board failure. The findings indicate that the strategy improves the evaluation markers, including the early warning accuracy, precision, recall, and F1 score, by an average of more than 10%. There is a 35% improvement in the computing efficiency, and the model can provide feature causal graph verification for expert product decision-making.

Causal graph extraction from news: a comparative study of time-series causality learning techniques.

Causal graph extraction from news has the potential to aid in the understanding of complex scenarios. In particular, it can help explain and predict events, as well as conjecture about possible cause-effect connections. However, limited work has addressed the problem of large-scale extraction of causal graphs from news articles. This article presents a novel framework for extracting causal graphs from digital text media. The framework relies on topic-relevant variables representing terms and ongoing events that are selected from a domain under analysis by applying specially developed information retrieval and natural language processing methods. Events are represented as event-phrase embeddings, which make it possible to group similar events into semantically cohesive clusters. A time series of the selected variables is given as input to a causal structure learning techniques to learn a causal graph associated with the topic that is being examined. The complete framework is applied to the New York Times dataset, which covers news for a period of 246 months (roughly 20 years), and is illustrated through a case study. An initial evaluation based on synthetic data is carried out to gain insight into the most effective time-series causality learning techniques. This evaluation comprises a systematic analysis of nine state-of-the-art causal structure learning techniques and two novel ensemble methods derived from the most effective techniques. Subsequently, the complete framework based on the most promising causal structure learning technique is evaluated with domain experts in a real-world scenario through the use of the presented case study. The proposed analysis offers valuable insights into the problems of identifying topic-relevant variables from large volumes of news and learning causal graphs from time series.

Generalized Gaussian quasi-maximum likelihood estimation for most common time series

We propose a consistent estimator for the parameter shape of the generalized gaussian noise in the class of causal time series including ARMA, AR(∞), GARCH, ARCH(∞), ARMA-GARCH, APARCH, ARMA-APARCH,…, processes. As well we prove the consistency and the asymptotic normality of the Generalized Gaussian Quasi-Maximum Likelihood Estimator (GGQMLE) for this class of causal time series with any fixed parameter shape, which over-performs the efficiency of the classical Gaussian QMLE. Monte Carlo experiments confirm that the accuracy of the proposed estimators.

Could Social Bots’ Sentiment Engagement Shape Humans’ Sentiment on COVID-19 Vaccine Discussion on Twitter?

During the COVID-19 pandemic, social media has become an emerging platform for the public to find information, share opinions, and seek coping strategies. Vaccination, one of the most effective public health interventions to control the COVID-19 pandemic, has become the focus of public online discussions. Several studies have demonstrated that social bots actively involved in topic discussions on social media and expressed their sentiments and emotions, which affected human users. However, it is unclear whether social bots’ sentiments affect human users’ sentiments of COVID-19 vaccines. This study seeks to scrutinize whether the sentiments of social bots affect human users’ sentiments of COVID-19 vaccines. The work identified social bots and built an innovative computational framework, i.e., the BERT-CNN sentiment analysis framework, to classify tweet sentiments at the three most discussed stages of COVID-19 vaccines on Twitter from December 2020 to August 2021, thus exploring the impacts of social bots on online vaccine sentiments of humans. Then, the Granger causality test was used to analyze whether there was a time-series causality between the sentiments of social bots and humans. The findings revealed that social bots can influence human sentiments about COVID-19 vaccines. Their ability to transmit the sentiments on social media, whether in the spread of positive or negative tweets, will have a corresponding impact on human sentiments.

Application of Hidden Markov Model in Financial Time Series Data

Financial time series have typical characteristics such as outliers, trends, and mean reversion. The existence of outliers will affect the effectiveness of the unknown parameter estimation in the financial time series forecasting model, so that the forecasting error of the model will be larger. Quantitative forecasting methods are divided into causal forecasting method and time series forecasting method. The causal forecasting method uses the causal relationship between the predictor variable and other variables to predict, and the time series forecasting method infers the future value of the predictor variable based on the structure of the historical data of the predictor. Therefore, this paper proposes a hidden Markov model prediction method based on the observation vector sequence, which can simultaneously consider the influence of the variable sequence structure and related factors.

DIC-ST: A Hybrid Prediction Framework Based on Causal Structure Learning for Cellular Traffic and Its Application in Urban Computing

The development of technology has strongly affected regional urbanization. With development of mobile communication technology, intelligent devices have become increasingly widely used in people’s lives. The application of big data in urban computing is multidimensional; it has been involved in different fields, such as urban planning, network optimization, intelligent transportation, energy consumption and so on. Data analysis becomes particularly important for wireless networks. In this paper, a method for analyzing cellular traffic data was proposed. Firstly, a method to extract trend components, periodic components and essential components from complex traffic time series was proposed. Secondly, we introduced causality data mining. Different from traditional time series causality analysis, the depth of causal mining was increased. We conducted causality verification on different components of time series and the results showed that the causal relationship between base stations is different in trend component, periodic component and essential component in urban wireless network. This is crucial for urban planning and network management. Thirdly, DIC-ST: a spatial temporal time series prediction based on decomposition and integration system with causal structure learning was proposed by combining GCN. Final results showed that the proposed method significantly improves the accuracy of cellular traffic prediction. At the same time, this method can play a crucial role for urban computing in network management, intelligent transportation, base station siting and energy consumption when combined with remote sensing map information.

Epidemic change-point detection in general causal time series

We consider an epidemic change-point detection in a large class of causal time series models, including among other processes, AR(∞), ARCH(∞), TARCH(∞), ARMA-GARCH. A test statistic based on the Gaussian quasi-maximum likelihood estimator of the parameter is proposed. It is shown that, under the null hypothesis of no change, the test statistic converges to a distribution obtained from a difference of two Brownian bridge and diverges to infinity under the epidemic alternative. Numerical results for simulation and real data example are provided.

THE ROLE OF INVESTOR SENTIMENT ON PRICING EFFECT OF PROFITABILITY RISK FACTOR IN KENYA

Emerging stock markets are characterized by strong investor sentiment and rapid fluctuations in returns. However, the role of investor sentiment on asset pricing has not been explored in these markets. This study sought to establish if the effect of the profitability risk factor on stock returns would vary with the level of investor sentiment at the Kenyan equity market. A quantitative causal time-series design was adopted to analyze the cause-effect relationship among the study variables. The study utilized monthly equity return data on 60 firms listed at the Nairobi Securities Exchange (NSE) from 2011 to 2019. Test portfolios were constructed following the Fama-French five-factor model framework. Auto-Regressive Distributed Lag (ARDL) and Vector Error Correction (VEC) estimation techniques show that profitability risk factor is a significant predictor of stock returns at a 5% level. Overall, though not consistent with valuation theory, the coefficient on profitability risk factor is negative, implying a high exposure to profitability risk results in low returns. Further, adding sentiment variables to the main effects model would enhance the significance of the profitability risk factor at the NSE. The evidence presented contributes to establishing investment strategies, estimating the required rate of return and assessing portfolio performance of collective investments.

Inference and model selection in general causal time series with exogenous covariates

In this paper, we study a general class of causal processes with exogenous covariates, including many classical processes such as the ARMA-GARCH, APARCH, ARMAX, GARCH-X and APARCH-X processes. Under some Lipschitz-type conditions, the existence of a τ-weakly dependent strictly stationary and ergodic solution is established. We provide conditions for the strong consistency and derive the asymptotic distribution of the quasi-maximum likelihood estimator (QMLE), both when the true parameter is an interior point of the parameters space and when it belongs to the boundary. A significance Wald-type test of parameter is developed. This test is quite extensive and includes the test of nullity of the parameter’s components, which in particular, allows us to assess the relevance of the exogenous covariates. Relying on the QMLE of the model, we also propose a penalized criterion to address the problem of the model selection for this class. The weak and the strong consistency of the procedure are established. Finally, Monte Carlo simulations are conducted to numerically illustrate the main results.