Noncausal Models Research Articles

Time aggregation of mixed causal–noncausal models

We study systematic and flow aggregation of mixed causal-noncausal autoregressive models. We show that aggregation preserves noncausality and generates a moving average component. Monte Carlo simulations demonstrate that backward- and forward-looking behavior can be identified empirically for sufficiently large samples.

Insomnia and short sleep duration, but not chronotype, is associated with chronic widespread pain: Mendelian randomization study.

Sleep disturbance has been associated with chronic widespread pain (CWP), but their causal relationship remains unclear. We aimed to examine the causal relationship and direction between CWP and sleep traits, namely insomnia, sleep duration and chronotype, using Mendelian Randomization. We used genetic association data from ~0.5 million individuals and up to 1.8 million controls from the UK Biobank (UKB). All traits were defined predominantly by self-report. Short sleep duration was defined as average ≤6 hours per 24 hours. Chronotype refers to the inclination to sleep at certain times where some wake and go to bed early ('morning' person), and others wake and go to sleep later ('evening' person). To permit use of the largest available genetic association data, we used the Causal Analysis Using Summary Effect estimates (CAUSE) method, which allows for sample overlap. Insomnia (OR 1.009, 95% credible interval 1.005, 1.014; p = 0.018 that the causal model is a better fit than non-causal model) and short sleep duration (OR 1.060, 95%CrI 1.038, 1.083; p = 0.040) were causally associated with increased risk of CWP, with limited evidence for reverse causation. There was no evidence in support of long sleep duration or chronotype being associated with CWP. This study suggest that insomnia and short sleep duration (≤6 hours) are associated with an increased risk of CWP. Improving short sleep duration and insomnia, rather than chronotype, may be effective in reducing the risk of CWP, although these results should be replicated in epidemiological and interventional studies.

Optimization of the generalized covariance estimator in noncausal processes

This paper investigates the performance of routinely used optimization algorithms in application to the Generalized Covariance estimator (GCov) for univariate and multivariate mixed causal and noncausal models. The GCov is a semi-parametric estimator with an objective function based on nonlinear autocovariances to identify causal and noncausal orders. When the number and type of nonlinear autocovariances included in the objective function are insufficient/inadequate, or the error density is too close to the Gaussian, identification issues can arise. These issues result in local minima in the objective function, which correspond to parameter values associated with incorrect causal and noncausal orders. Then, depending on the starting point and the optimization algorithm employed, the algorithm can converge to a local minimum. The paper proposes the Simulated Annealing (SA) optimization algorithm as an alternative to conventional numerical optimization methods. The results demonstrate that SA performs well in its application to mixed causal and noncausal models, successfully eliminating the effects of local minima. The proposed approach is illustrated by an empirical study of a bivariate series of commodity prices.

Making predictions under interventions: a case study from the PREDICT-CVD cohort in New Zealand primary care.

Most existing clinical prediction models do not allow predictions under interventions. Such predictions allow predicted risk under different proposed strategies to be compared and are therefore useful to support clinical decision making. We aimed to compare methodological approaches for predicting individual level cardiovascular risk under three interventions: smoking cessation, reducing blood pressure, and reducing cholesterol. We used data from the PREDICT prospective cohort study in New Zealand to calculate cardiovascular risk in a primary care setting. We compared three strategies to estimate absolute risk under intervention: (a) conditioning on hypothetical interventions in non-causal models; (b) combining existing prediction models with causal effects estimated using observational causal inference methods; and (c) combining existing prediction models with causal effects reported in published literature. The median absolute cardiovascular risk among smokers was 3.9%; our approaches predicted that smoking cessation reduced this to a median between a non-causal estimate of 2.5% and a causal estimate of 2.8%, depending on estimation methods. For reducing blood pressure, the proposed approaches estimated a reduction of absolute risk from a median of 4.9% to a median between 3.2% and 4.5% (both derived from causal estimation). Reducing cholesterol was estimated to reduce median absolute risk from 3.1% to between 2.2% (non-causal estimate) and 2.8% (causal estimate). Estimated absolute risk reductions based on non-causal methods were different to those based on causal methods, and there was substantial variation in estimates within the causal methods. Researchers wishing to estimate risk under intervention should be explicit about their causal modelling assumptions and conduct sensitivity analysis by considering a range of possible approaches.

Feature selection in text classification: Identifying spurious words with causal inference methods

As has been scrutinized by many, non-causal model may contain spurious correlations that act like shortcuts during the prediction phase, undermining cross-domain accuracy. This can be caused by biased training data that contains spurious words with neutral meanings yet can induce the model to predict wrongly. Based on this assumption, we propose a series of methods to detect these spurious words before feeding the model with the training data. We used advanced causal inference methods which are arising novas in recent studies, such as propensity score matching and inverse propensity score weighting to facilitate the feature selection before training. We experimented with multiple approaches to estimate propensity scores and got profound improvements. We further experimented with BERT model to evaluate the effectiveness of feature selection and find that the model performance with in-domain and out-of-domain testing samples is boosted after we remove the spurious words detected by our methods in the training data.

On the Different Abilities of Cross-Sample Entropy and K-Nearest-Neighbor Cross-Unpredictability in Assessing Dynamic Cardiorespiratory and Cerebrovascular Interactions.

Nonlinear markers of coupling strength are often utilized to typify cardiorespiratory and cerebrovascular regulations. The computation of these indices requires techniques describing nonlinear interactions between respiration (R) and heart period (HP) and between mean arterial pressure (MAP) and mean cerebral blood velocity (MCBv). We compared two model-free methods for the assessment of dynamic HP-R and MCBv-MAP interactions, namely the cross-sample entropy (CSampEn) and k-nearest-neighbor cross-unpredictability (KNNCUP). Comparison was carried out first over simulations generated by linear and nonlinear unidirectional causal, bidirectional linear causal, and lag-zero linear noncausal models, and then over experimental data acquired from 19 subjects at supine rest during spontaneous breathing and controlled respiration at 10, 15, and 20 breaths·minute-1 as well as from 13 subjects at supine rest and during 60° head-up tilt. Linear markers were computed for comparison. We found that: (i) over simulations, CSampEn and KNNCUP exhibit different abilities in evaluating coupling strength; (ii) KNNCUP is more reliable than CSampEn when interactions occur according to a causal structure, while performances are similar in noncausal models; (iii) in healthy subjects, KNNCUP is more powerful in characterizing cardiorespiratory and cerebrovascular variability interactions than CSampEn and linear markers. We recommend KNNCUP for quantifying cardiorespiratory and cerebrovascular coupling.

Disentangling the relationship between depression and chronic widespread pain: A Mendelian randomisation study

ObjectiveDepression and chronic widespread pain (CWP) frequently coexist, but whether depression is an independent causal risk factor for CWP, and/or vice versa, remains unclear. We investigated the bidirectional causal relationship between depression and CWP. MethodsWe performed a two‐sample Mendelian randomisation (MR) study to estimate the causal relationship between genetically predicted depression (170,756 cases, 329,443 controls) and risk of CWP (6,914 cases, 242,929 controls), and the effect of CWP on depression susceptibility, using large population-level genetic data. We used a new MR method, Causal Analysis Using Summary Effect estimates (CAUSE), which allows for sample overlap, in addition to traditional MR and sensitivity analyses. ResultsFor each doubling in odds of genetic liability to depression, the risk of chronic widespread pain was increased (OR 1.004, 95% credible interval 1.003–1.005; p = 7.3 × 10−5 that the causal model is a better fit than non-causal model). There was bidirectional evidence of causality, with genetic liability to chronic widespread pain increasing depression susceptibility (OR 2.31; 95%CrI 1.57, 3.40; p = 0.0026 that the causal model is a better fit). Other MR methods produced concordant results. ConclusionsThis study provides evidence in support of a bidirectional causal relationship between depression and increased risk of chronic widespread pain, whilst overcoming the major limitations of previous epidemiological studies. Interventions for depression may be an effective strategy to prevent or reduce the burden of chronic widespread pain and vice versa.

Detecting Common Bubbles in Multivariate Mixed Causal–Noncausal Models

This paper proposes concepts and methods to investigate whether the bubble patterns observed in individual time series are common among them. Having established the conditions under which common bubbles are present within the class of mixed causal–noncausal vector autoregressive models, we suggest statistical tools to detect the common locally explosive dynamics in a Student t-distribution maximum likelihood framework. The performances of both likelihood ratio tests and information criteria were investigated in a Monte Carlo study. Finally, we evaluated the practical value of our approach via an empirical application on three commodity prices.

Gender and sex bias in COVID-19 epidemiological data through the lens of causality

The COVID-19 pandemic has spurred a large amount of experimental and observational studies reporting clear correlation between the risk of developing severe COVID-19 (or dying from it) and whether the individual is male or female. This paper is an attempt to explain the supposed male vulnerability to COVID-19 using a causal approach. We proceed by identifying a set of confounding and mediating factors, based on the review of epidemiological literature and analysis of sex-dis-aggregated data. Those factors are then taken into consideration to produce explainable and fair prediction and decision models from observational data. The paper outlines how non-causal models can motivate discriminatory policies such as biased allocation of the limited resources in intensive care units (ICUs). The objective is to anticipate and avoid disparate impact and discrimination, by considering causal knowledge and causal-based techniques to compliment the collection and analysis of observational big-data. The hope is to contribute to more careful use of health related information access systems for developing fair and robust predictive models.

Is Climate Change Time-Reversible?

This paper proposes strategies to detect time reversibility in stationary stochastic processes by using the properties of mixed causal and noncausal models. It shows that they can also be used for non-stationary processes when the trend component is computed with the Hodrick–Prescott filter rendering a time-reversible closed-form solution. This paper also links the concept of an environmental tipping point to the statistical property of time irreversibility and assesses fourteen climate indicators. We find evidence of time irreversibility in greenhouse gas emissions, global temperature, global sea levels, sea ice area, and some natural oscillation indices. While not conclusive, our findings urge the implementation of correction policies to avoid the worst consequences of climate change and not miss the opportunity window, which might still be available, despite closing quickly.

Causal Decision Making and Causal Effect Estimation Are Not the Same…and Why It Matters

Causal decision making (CDM) at scale has become a routine part of business, and increasingly, CDM is based on statistical models and machine learning algorithms. Businesses algorithmically target offers, incentives, and recommendations to affect consumer behavior. Recently, we have seen an acceleration of research related to CDM and causal effect estimation (CEE) using machine-learned models. This article highlights an important perspective: CDM is not the same as CEE, and counterintuitively, accurate CEE is not necessary for accurate CDM. Our experience is that this is not well understood by practitioners or most researchers. Technically, the estimand of interest is different, and this has important implications both for modeling and for the use of statistical models for CDM. We draw on recent research to highlight three implications. (1) We should carefully consider the objective function of the causal machine learning, and if possible, optimize for accurate “treatment assignment” rather than for accurate effect-size estimation. (2) Confounding affects CDM and CEE differently. The upshot here is that for supporting CDM it may be just as good or even better to learn with confounded data as with unconfounded data. (3) Causal statistical modeling may not be necessary at all to support CDM because a proxy target for statistical modeling might do as well or better. This third observation helps to explain at least one broad common CDM practice that seems “wrong” at first blush—the widespread use of noncausal models for targeting interventions. The last two implications are particularly important in practice, as acquiring (unconfounded) data on both “sides” of the counterfactual for modeling can be quite costly and often impracticable. These observations open substantial research ground. We hope to facilitate research in this area by pointing to related articles from multiple contributing fields, most of them written in the last five years.

Inference in mixed causal and noncausal models with generalized Student’s t-distributions

The properties of Maximum Likelihood estimator in mixed causal and noncausal models with a generalized Student’s t error process are reviewed. Several known existing methods are typically not applicable in the heavy-tailed framework. To this end, a new approach to make inference on causal and noncausal parameters in finite sample sizes is proposed. It exploits the empirical variance of the generalized Student’s t, without the existence of population variance. Monte Carlo simulations show a good performance of the new variance construction for fat tail series. Finally, different existing approaches are compared using three empirical applications: the variation of daily COVID-19 deaths in Belgium, the monthly wheat prices, and the monthly inflation rate in Brazil.

A time-varying parameter model for local explosions

Financial and economic time series can feature locally explosive behaviour when bubbles are formed. We develop a time-varying parameter model that is capable of describing this behaviour in time series data. Our proposed dynamic model can be used to predict the emergence, existence and burst of bubbles. We adopt a flexible observation driven model specification that allows for different bubble shapes and behaviour. We establish stationarity, ergodicity, and bounded moments of the data generated by our model. Furthermore, we obtain the consistency and asymptotic normality of the maximum likelihood estimator. Given the parameter estimates in the model, the implied filter is capable of extracting the unobserved bubble process from the observed data. We study finite-sample properties of our estimator through a Monte Carlo simulation study. Finally, we show that our model compares well with existing noncausal models in a financial application concerning the Bitcoin/US dollar exchange rate.

Modular Compilation for a Hybrid Non-Causal Modelling Language

Non-causal modelling is a powerful approach to modelling physical systems in a variety of domains from science and engineering. Non-causal modelling languages enable a high-level and modular approach to modelling. However, it is hard to compile non-causal languages modularly (in the sense of separate compilation). This causes difficulties when simulating large models for which code generation takes a long time, or structurally singular models in which parts of the model are allowed to change at runtime. In this work, we introduce a technique we call order-parametric differentiation to allow truly modular compilation. The idea is to generate (machine) code that can compute derivatives of any order of an expression as needed, thus allowing for ahead-of-time modular compilation of a hybrid non-causal language. We also develop a compilation scheme that enables using partial models as first-class objects in a seamless way and simulating them without the need for just-in-time compilation, even in the presence of structural dynamism. We present a performance evaluation of the scheme we used and study its shortcomings and possible improvements, demonstrating that it is a feasible complement to existing implementation techniques for cases where true modular compilation is a primary objective.

Selecting between causal and noncausal models with quantile autoregressions

Abstract We propose a model selection criterion to detect purely causal from purely noncausal models in the framework of quantile autoregressions (QAR). We also present asymptotics for the i.i.d. case with regularly varying distributed innovations in QAR. This new modelling perspective is appealing for investigating the presence of bubbles in economic and financial time series, and is an alternative to approximate maximum likelihood methods. We illustrate our analysis using hyperinflation episodes of Latin American countries.

Learning from Non-Causal Models

This paper defends the thesis of learning from non-causal models: viz. that the study of some model can prompt justified changes in one’s confidence in empirical hypotheses about a real-world target in the absence of any known or predicted similarity between model and target with regards to their causal features. Recognizing that we can learn from non-causal models matters not only to our understanding of past scientific achievements, but also to contemporary debates in the philosophy of science. At one end of the philosophical spectrum, my thesis undermines the views of those who, like Cartwright (Erkenntnis 70:45–58, 2009), follow Hesse (Models and Analogies in Science, Notre Dame, University of Notre Dame Press, 1963) in restricting the possibility of learning from models to only those situations where a model identifies some causal factors present in the target. At the other end of the spectrum, my thesis also helps undermine some extremely permissive positions, e.g., Grüne-Yanoff’s (Erkenntnis 70(1):81–99, 2009, Philos Sci 80(5): 850–861, 2013) claim that learning from a model is possible even in the absence of any similarity at all between model and target. The thesis that we can learn from non-causal models offers a cautious middle ground between these two extremes.

Dynamic Modeling and Heat Flow Study of a Thermal Power Plant Using OpenModelica

With the integration of intermittent energy sources in the power system, the main focus of thermal power plant design is to optimize the conventional system by improving efficiency and flexibility of power generation in the start-up and base-load procedures. This paper focuses on the behavior of thermal power generation with respect to load variations, fuel variations, weather conditions and internal dynamics of the system to aid in energy planning and the optimization of power system. Dynamic modeling and subsequent simulation is to make improvements. Therefore, to model the system, this research study has used a dynamic modeling and a simulation framework (OpenModelica based on modelica language) which is an object-oriented and equation-based language. This framework allows a non-causal modeling and a flexible reuse of the models since data-flow direction is not predefined like other state of the art simulating tools i.e. MATLAB. The proposed model analyses the trend of electrical power generation of KOT ADDU thermal power plant to predict the future demand of electrical power. The achieved results show more than 90% accuracy of the proposed model in the transient operation such as start-up and base load procedure.

Explanations and candidate explanations in physics

There has been a growing trend to include non-causal models in accounts of scientific explanation. A worry addressed in this paper is that without a higher threshold for explanation there are no tools for distinguishing between models that provide genuine explanations and those that provide merely potential explanations. To remedy this, a condition is introduced that extends a veridicality requirement to models that are empirically underdetermined, highly-idealised, or otherwise non-causal. This condition is applied to models of electroweak symmetry breaking beyond the Standard Model.

Identification of symmetric noncausal processes

We propose a maximum likelihood estimation approach for the identification of symmetric noncausal models. Such models are used to represent the cross-directional dynamic response of many industrial processes that are generally modeled with a high-dimensional gain matrix. Reducing the number of parameters in a noncausal model can significantly reduce the uncertainty associated with parameter estimates. We adapt the maximum likelihood method to treat symmetric noncausal models by showing that every symmetric noncausal process admits a spectrally equivalent causal model. It is then proved that the maximum likelihood estimate of this causal model converges to that of the original noncausal model. We present an iterative identification algorithm to efficiently estimate the parameters in noncausal models. Finally, we show that the parameter covariance estimate obtained from the causal model also converges to that of the noncausal model, which lays a foundation for optimal input design in noncausal processes. Several numerical examples illustrate the effectiveness of the proposed algorithm.

Feedback noncausal model predictive control of wave energy converters

In this paper, a novel feedback noncausal model predictive control (MPC) strategy for sea wave energy converters (WECs) is proposed, where the wave prediction information can be explicitly incorporated into the MPC strategy to improve the WEC control performance. The main novelties of the MPC strategy proposed in this paper include: (i) the recursive feasibility and robust constraints satisfaction are guaranteed without a significant increase in the computational burden; (ii) the information of short-term wave prediction is incorporated into the feedback noncausal MPC method to maximise the potential energy output; (iii) the sea condition for the WEC to safely operate in can be explicitly calculated. The proposed feedback noncausal MPC algorithm can also be extended to a wide class of control design problems, especially to the energy maximisation problems with constraints to be satisfied and subject to persistent but predictable disturbances. Numerical simulations are provided to show the efficacy of the proposed feedback noncausal MPC.