Data-driven Sensitivity Analysis Research Articles

Unveiling the Sensitivity Analysis of Port Carbon Footprint via Power Alternative Scenarios: A Deep Dive into the Valencia Port Case Study

The Port of Valencia, a prominent maritime center, is actively working towards minimizing its carbon emissions and aims to become a completely carbon-neutral port soon. This research uses data-driven sensitivity analysis to explore realistic power-generating options for a seaport to reduce its emissions. This approach comprises changing key parameters in power consumption and deploying renewable energies (rather than electricity and infrastructure prices, which are beyond the scope of this study) to assess their impact on the port’s overall emissions profile. Through sensitivity analysis, policymakers and managers discover each scenario’s efficacy and find the best decarbonization strategies. After thoroughly examining four realistic scenarios, our research findings show that each scenario’s emission reduction share and sensitivity are practical and feasible. It becomes clear that gradually replacing traditional fossil fuels for electricity generation with renewables is a reasonable and realistic option for emissions reduction. The results demonstrate that focusing on reasonable targets, such as replacing 30% and 50% of electricity generation with renewables, is more achievable and beneficial in the medium term than ambitious goals, like replacing all electricity with renewable energy. This research contributes to reducing emissions of the Port of Valencia by using data-driven sensitivity analysis to find practical renewable energy strategies. It provides actionable insights for managers and policymakers to implement feasible decarbonization plans, emphasizing gradual adoption of renewables over ambitious goals, thus supporting sustainable maritime operations.

Performance of flooded flexible pavements: A data-driven sensitivity analysis considering soil moisture fluctuations

The design of pavement systems relies on sensitivity of their performance to moisture variations; yet traditional sensitivity analysis have limitations in understanding the complex interplays affecting pavement performance, and they can be computationally expensive. This study introduces a novel approach of employing System Dynamics Models (SDM) to assess the factors influencing flexible pavement performance under moisture variations. A data-driven Global Sensitivity Analysis (GSA) can complement traditional methods such as Local Sensitivity Analysis (LSA) and feature ranking, providing a deeper understanding of interconnections. Utilizing Latin Hypercube Sampling, input variable samples are generated to facilitate data-driven GSA. To balance prediction accuracy with computational efficiency, a Random Forest ensemble is developed as a multitarget surrogate model. Key findings underscore the interplay between resilience modulus and pavement layer thickness as well as the significance of subgrade soil particle size for coarse-grained subgrades, while subbase aggregate properties show negligible influence. In fine-grained subgrades, soil plasticity index plays a vital role in peak surface deflection. Additionally, the study uncovers the critical impact of precipitation variables on deflection timing. By predicting these critical timings, informed traffic management measures can be contemplated, enhancing road user safety. This research highlights the importance of tailoring pavement design strategies to specific subgrade characteristics and environmental conditions. Moreover, the study recommends adoption of the proposed sensitivity analysis workflow to delve into the various parameters interactions in infrastructure system design.

A machine learning approach to study randomness in enhanced tensile properties of KOH-treated jute fabric

The probabilistic characterization of tensile strength of potassium hydroxide (KOH) treated jute fiber is conducted in the light of machine learning methods. The concentration of KOH, soaking periodand drying temperature are considered as the input features. The experimental results after surface treatment with KOHare utilized to determine the machine learning models. The predictive capabilities of linear regression (LR) and Gaussian process regression (GPR) based modelsare compared and it has been found that the latter is more accurate than the former. Further, the data-driven sensitivity analysis and uncertainty quantification arecarried out with the help of GPRmodel. The overall probabilistic results reported in this articlewould provide the complete insight regarding tensile strength of potassium hydroxide (KOH) treated jute under the uncertain variations in the process parameters.Thestudy also provides the total design domain of treated jute under the practically relevant uncertainty associated with the discussed process parameters.

Probabilistic investigation of geometric responses in Wire EDM machined complex-shaped profile: A machine learning based approach

Wire electric discharge machining (WEDM) has gained tremendous market share due to its potential to create complex profiles and ease in machining exotic materials. The inherent problem of wire lag, however, has a significant impact on the accuracy and precision of complex profiles in WEDM. Furthermore, if parametric uncertainties are not taken into account during machining of complex profiles, problems like geometrical inaccuracy and imprecision are likely to worsen. Hence, in the present study, we focused on the role of parametric uncertainties on geometrical parameters such as corner error (CE) and undercut (UC) of complicated profile. In this regard, the machine learning (ML) driven framework is proposed, wherein the Gaussian process regression model is integrated with the experimental responses gathered from the WEDM. The sample-space to train and validate the ML model is constructed by performing the WEDM for the experimental plan, which is designed on the basis of face-centered central composite design. The constructed ML model is rigorously tested and validated to ensure its predictive accuracy. The ML-based WEDM framework is further extended to perform data-driven uncertainty quantification and sensitivity analysis to reveal the likelihood of variations in geometrical precision due to inherent parametric uncertainties. The constructed model is further deployed for the data-driven sensitivity analysis and to reveal the probabilistic behavior of responses in the presence of parametric uncertainties. The results of this study will help to develop and expand the WEDM robust control technique.

Score rectification for online assessments in robot-assisted arm rehabilitation

AbstractRelative comparison of clinical scores to measure the effectiveness of neuro-rehabilitation therapy is possible through a series of discrete measurements during the rehabilitation period within specifically designed task environments. Robots allow quantitative, continuous measurement of data. Resulting robotic scores are also only comparable within similar context, e.g. type of task. We propose a method to decouple these scores from their respective context through functional orthogonalization and compensation of the compounding factors based on a data-driven sensitivity analysis of the user performance. The method was validated for the established accuracy score with variable arm weight support, provoked muscle fatigue and different task directions on 6 participants of our arm exoskeleton group on the ANYexo robot. In the best case, the standard deviation of the assessed score in changing context could be reduced by a factor of 3.2. Therewith, we paved the way to context-independent, quantitative online assessments, recorded autonomously with robots.

Data-Driven Predictive Maintenance for Gas Distribution Networks

A generic data-driven approach is presented that employs machine learning to predict the future reliability of components in utility networks. The proposed approach enables utilities to implement a predictive maintenance strategy that optimizes life-cycle costs without compromising safety or creating environmental issues. Any machine learning technique that qualifies as a probabilistic classifier can be employed within the proposed approach. To identify the data-driven model that performs best, a practical metric to assess the performance of the competing models is proposed. This metric is specifically designed to quantify the forecasting performance with respect to maintenance planning. Additionally, a data-driven sensitivity analysis approach is discussed that allows for an assessment of the influence of the different features on the model prediction. Through an application example, it is demonstrated how the proposed approach can be applied to predict future defect rates of pipe sections for maintenance planning in a large gas distribution network.

Machine Labor

The utility of machine learning (ML) for regression-based causal inference is illustrated by using lasso to select control variables for estimates of college characteristics’ wage effects. Post-double-selection lasso offers a path to data-driven sensitivity analysis. ML also seems useful for an instrumental variables (IV) first stage, since two-stage least squares (2SLS) bias reflects overfitting. While ML-based instrument selection can improve on 2SLS, split-sample IV and limited information maximum likelihood do better. Finally, we use ML to choose IV controls. Here, ML creates artificial exclusion restrictions, generating spurious findings. On balance, ML seems ill-suited to IV applications in labor economics.

Data-driven global sensitivity indices for the load-carrying capacity of large-diameter stiffened cylindrical shells

The influence of input parameters on the load-carrying capacity of the large-diameter stiffened cylindrical shell has not been satisfactory understood. To obtain the global sensitivity indices for the load-carrying capacity of large-diameter stiffened cylindrical shells, a novel data-driven sensitivity analysis method is presented for the efficient calculation of Sobol’s indices. In the work, the analytical expressions to compute Sobol’s indices are derived in detail based on the Radial Basis Function (RBF) metamodel. Then, the collapse mode of stiffened cylindrical shells with different sections of stringers are simulated, and the sensitivity of the load-carrying performance to geometric parameters are analyzed. The results show that for the considered stiffened cylindrical shell: (1) The coupled flexural-torsional induced deformation becomes the main factor leading to the overall collapse of the large-diameter and heavy-load stiffened cylindrical shells. (2) The flange thickness of stringers has the strongest influence on the load-bearing capacity, followed by the web thickness of stringers. (3) The interaction influences between stringers and other components in stiffened cylindrical shells are insignificant and negligible compared to the individual influence of stringers.

Searching for axionlike particles from core-collapse supernovae with Fermi LAT’s low-energy technique

Light axion-like particles (ALPs) are expected to be abundantly produced in core-collapse supernovae (CCSNe), resulting in a $\sim$10-second long burst of ALPs. These particles subsequently undergo conversion into gamma-rays in external magnetic fields to produce a long gamma-ray burst (GRB) with a characteristic spectrum peaking in the 30--100-MeV energy range. At the same time, CCSNe are invoked as progenitors of {\it ordinary} long GRBs, rendering it relevant to conduct a comprehensive search for ALP spectral signatures using the observations of long GRB with the \textit{Fermi} Large Area Telescope (LAT). We perform a data-driven sensitivity analysis to determine CCSN distances for which a detection of an ALP signal is possible with the LAT's low-energy (LLE) technique which, in contrast to the standard LAT analysis, allows for a a larger effective area for energies down to 30~MeV. Assuming an ALP mass $m_a \lesssim 10^{-10}$~eV and ALP-photon coupling $g_{a\gamma} = 5.3\times 10^{-12}$ GeV$^{-1}$, values considered and deduced in ALP searches from SN1987A, we find that the distance limit ranges from $\sim\!0.5$ to $\sim\!10$~Mpc, depending on the sky location and the CCSN progenitor mass. Furthermore, we select a candidate sample of twenty-four GRBs and carry out a model comparison analysis in which we consider different GRB spectral models with and without an ALP signal component. We find that the inclusion of an ALP contribution does not result in any statistically significant improvement of the fits to the data. We discuss the statistical method used in our analysis and the underlying physical assumptions, the feasibility of setting upper limits on the ALP-photon coupling, and give an outlook on future telescopes in the context of ALP searches.

Data-Driven Sensitivity Analysis for Static Mechanical Properties of Additively Manufactured Ti–6Al–4V

Abstract Additive manufacturing (AM) has been extensively investigated in recent years to explore its application in a wide range of engineering functionalities, such as mechanical, acoustic, thermal, and electrical properties. A data-driven approach is proposed to investigate the influence of major fabrication parameters in the laser-based additively manufactured Ti–6Al–4V. Two separate laser-based powder bed fusion techniques, i.e., selective laser melting (SLM) and direct metal laser sintering (DMLS), have been investigated and several data regarding the tensile properties of Ti–6Al–4V alloy with their corresponding fabrication parameters are collected from open literature. Statistical data analysis is performed for four fabrication parameters (scanning speed, laser power, hatch spacing, and powder layer thickness) and three postfabrication parameters (heating temperature, heating time, and hot isostatically pressed or not) which are major influencing factors and have been investigated by several researchers to identify their behavior on the static mechanical properties (i.e., yielding strength, ultimate tensile strength, and elongation). To identify the behavior of the relationship between the input and output parameters, both linear regression analysis and artificial neural network (ANN) models are developed using 53 and 100 datasets for SLM and DMLS processes, respectively. The linear regression model resulted in an average R squared value of 0.351 and 0.507 compared to 0.908 and 0.833 in the case of nonlinear ANN modeling for SLM and DMLS based modeling, respectively. Both local and global sensitivity analyses are carried out to identify the important factors for future optimal design. Based on the current study, local sensitivity analysis (SA) suggests that SLM is most sensitive to laser power, scanning speed, and heat treatment temperature while DMLS is most sensitive to heat treatment temperature, hatch spacing, and laser power. In the case of DMLS fabricated Ti–6Al–4V alloy, laser power, and scan speed are found to be the most impactful input parameters for tensile properties of the alloy while heating time turned out to be the least affecting parameter. The global sensitivity analysis results can be used to tailor the alloy's static properties as per the requirement while results from local sensitivity analysis could be useful to optimize the already tailored design properties. Sobol's global sensitivity analysis implicates laser power, heating temperature, and hatch spacing to be the most influential parameters for alloy strength while powder layer thickness followed by scanning speed to be the prominent parameters for elongation for SLM fabricated Ti–6Al–4V alloy. Future work would still be needed to eradicate some of the limitations of this study related to limited dataset availability.

Data-driven sensitivity analysis and electricity consumption prediction for water source heat pump system using limited information

The studies on predicting the energy consumption of air conditioning systems are meaningful to building energy conservation and management. Generally, the more comprehensive the building information is, the easier the prediction model can be developed. However, it is very difficult to get detailed information about existing/old buildings (information-poor buildings), it is a big challenge to predict the energy consumption accurately by limited information. This study aims to predict the electricity consumption of the water source heat pump system of an office building based on meteorological data. The key variables are selected by error analysis and sensitivity analysis, and the effects of each variable on the models’ prediction performance can be obtained. Besides, the prediction models are established by support vector regression algorithm and trained by the local meteorological data. The results show that the positive and negative variables can be identified, and these positive variables are responsible for more than 70% of the total importance. Moreover, the root mean square error falls to 4.6044 from 7.8227 and the relative square error falls to 0.1494 from 0.4313 when the negative inputs are removed. And the errors reduce further to 4.1160 and 0.1194 by parameter optimization.

A multiscale framework to predict electrochemical characteristics of yttrium doped Barium Zirconate based solid oxide cells

A multiscale model is developed combining physics at atomistic, meso and continuum scales to predict the electrochemical characteristics of Current-Voltage curves in Yttrium doped Barium Zirconate based solid oxide cells. The most probable reaction pathway involving proton transfer from the surface of electrodes to the electrode/air/electrolyte Triple Phase Boundary from where it moves to the electrolyte is proposed, and their reaction barriers have been predicted from Density Functional Theory calculations. The model is validated against experimental observations. The environment, such as the amount of H2 (reducing) and O2 (oxidizing) gases the electrolyte is equilibrated in, affects the type of charge carriers while the temperature affects their rate of transport. Both these factors affect the rate of reactions in the proposed pathway. These effects are well manifested in the predicted electrochemical characteristics. The reducing environments are suitable for fuel cell mode operation and, the oxidizing environments are suitable for electrolyzer cell mode operations. The activation energy for conductivity in oxidizing environments is higher making it less amenable for low-temperature operation than in reducing environments. A data-driven sensitivity analysis is performed from which a non-intuitive parameter, permittivity of the electrolyte, is predicted to be important for Solid Oxide Cell performance.

Data-driven sensitivity analysis of complex machine learning models: A case study of directional drilling

Classical sensitivity analysis of machine learning regression models is a topic sparse in literature. Most of data-driven models are complex black boxes with limited potential of extracting mathematical understanding of underlying model self-arranged through the training algorithm. Sensitivity analysis can uncover erratic behavior stemming from overfitting or insufficient size of the training dataset. It can also guide model evaluation and application. In this paper, our work on data-driven sensitivity analysis of complex machine learning models is presented. Rooted in one-at-a-time method it utilizes training, validation and testing datasets to cover the hyperspace of potential inputs. The method is highly scalable, it allows for sensitivity analysis of individual as well as groups of inputs. The method is not computationally expensive, scaling linearly both with the available data samples, and in relation to the quantity of inputs and outputs. Coupled with the fact that calculations are considered embarrassingly parallel, it makes the method attractive for big models. In the case study, a regression model to predict inclinations using recurrent neural network was employed to illustrate our proposed sensitivity analysis method and results.

Data-driven sensitivity analysis of contact resistance to assembly errors for proton-exchange membrane fuel cells

The proton-exchange membrane fuel cell is a promising power source for automobile industry because of its zero pollution. However, its stack structure always faces increased contact resistance caused by assembly errors, leading to substantial energy loss during the working period. To enhance its output performance, the influence of assembly errors on contact resistance is studied for proton-exchange membrane fuel cell. The mechanical simulation model of fuel cell assembly process is established to provide contact resistance distribution with different assembly errors. An improved global sensitivity analysis method is proposed to evaluate the influence coefficient of each assembly error term on contact resistance based on a series of randomized simulation data. The case study of a single-layer fuel cell demonstrates the proposed method achieves higher efficiency than traditional sensitivity analysis methods, and finds out key assembly errors in regard to reducing contact resistance.

Survivor-Complier Effects in the Presence of Selection on Treatment, With Application to a Study of Prompt ICU Admission

ABSTRACTPretreatment selection or censoring (“selection on treatment”) can occur when two treatment levels are compared ignoring the third option of neither treatment, in “censoring by death” settings where treatment is only defined for those who survive long enough to receive it, or in general in studies where the treatment is only defined for a subset of the population. Unfortunately, the standard instrumental variable (IV) estimand is not defined in the presence of such selection, so we consider estimating a new survivor-complier causal effect. Although this effect is generally not identified under standard IV assumptions, it is possible to construct sharp bounds. We derive these bounds and give a corresponding data-driven sensitivity analysis, along with nonparametric yet efficient estimation methods. Importantly, our approach allows for high-dimensional confounding adjustment, and valid inference even after employing machine learning. Incorporating covariates can tighten bounds dramatically, especially when they are strong predictors of the selection process. We apply the methods in a UK cohort study of critical care patients to examine the mortality effects of prompt admission to the intensive care unit, using ICU bed availability as an instrument. Supplementary materials for this article are available online.

Data-driven sensitivity analysis to detect missing data mechanism with applications to structural equation modelling

Missing data are a common problem in almost all areas of empirical research. Ignoring the missing data mechanism, especially when data are missing not at random (MNAR), can result in biased and/or inefficient inference. Because MNAR mechanism is not verifiable based on the observed data, sensitivity analysis is often used to assess it. Current sensitivity analysis methods primarily assume a model for the response mechanism in conjunction with a measurement model and examine sensitivity to missing data mechanism via the parameters of the response model. Recently, Jamshidian and Mata (Post-modelling sensitivity analysis to detect the effect of missing data mechanism, Multivariate Behav. Res. 43 (2008), pp. 432–452) introduced a new method of sensitivity analysis that does not require the difficult task of modelling the missing data mechanism. In this method, a single measurement model is fitted to all of the data and to a sub-sample of the data. Discrepancy in the parameter estimates obtained from the the two data sets is used as a measure of sensitivity to missing data mechanism. Jamshidian and Mata describe their method mainly in the context of detecting data that are missing completely at random (MCAR). They used a bootstrap type method, that relies on heuristic input from the researcher, to test for the discrepancy of the parameter estimates. Instead of using bootstrap, the current article obtains confidence interval for parameter differences on two samples based on an asymptotic approximation. Because it does not use bootstrap, the developed procedure avoids likely convergence problems with the bootstrap methods. It does not require heuristic input from the researcher and can be readily implemented in statistical software. The article also discusses methods of obtaining sub-samples that may be used to test missing at random in addition to MCAR. An application of the developed procedure to a real data set, from the first wave of an ongoing longitudinal study on aging, is presented. Simulation studies are performed as well, using two methods of missing data generation, which show promise for the proposed sensitivity method. One method of missing data generation is also new and interesting in its own right.