Search Results (268)

Precipitation records from meteorological stations frequently contain gaps caused by sensor, power, or transmission failures, creating uncertainty in hydrological, agricultural, and water-resources applications. This study compared two simple baselines (station-specific monthly climatological mean and temporal linear interpolation), deterministic spatial interpolation, direct reanalysis-based replacement, and machine-learning methods for daily precipitation imputation. Daily precipitation from 14 stations in Eastern and Southeastern Türkiye during 1985–2014 was evaluated using an independent final-test set formed by stratified random masking of 15% of complete observations; the remaining 85% was used for calibration, SHapley Additive exPlanations (SHAP) analysis, cross-validation, and hyperparameter optimization. ERA5-Land variables were transferred to the stations, precipitation was calibrated by Empirical Quantile Mapping, and leakage-controlled Kriging estimates were incorporated as predictors in XGBoost, LightGBM, Random Forest, Support Vector Regression, and Multilayer Perceptron models. The station-month climatological mean (RMSE = 5.4820 mm; NSE = 0.0527) and temporal linear interpolation (RMSE = 5.7059 mm; NSE = −0.0262) performed substantially worse than optimized Kriging and IDW. The full-hybrid LightGBM model achieved the best performance (RMSE = 3.2001 mm; MAE = 0.9814 mm; Pearson r = 0.8317; NSE = 0.6772), whereas direct ERA5-EQM replacement was less accurate (RMSE = 5.2252 mm; NSE = 0.1394). Combining local observations, spatial information, and ERA5-Land covariates therefore improved daily precipitation imputation in the study region. Full article

The impact of particulate matter (PM) on pediatric sleep has been documented, yet the roles of particle size and particulate constituents remain poorly understood. This study aims to evaluate the associations between early-life exposure to different sizes of PM and specific chemical constituents of PM_2.5 and sleep in toddlers. In this birth cohort in Guangzhou, China, 514 mother–child pairs with follow-ups at age two between 2017 and 2018 were eligible for participation. Using high-resolution ground-level grid data, we estimated PM exposure (PM₁, PM_2.5, PM₁₀) and major chemical constituents of PM_2.5 during the early-life period. The sleep of two-year-old toddlers was assessed using the Chinese version of the Brief Infant Sleep Questionnaire. We used multivariable linear and logistic regression models to analyze the associations between PM exposure and sleep. Quantile g-computation was applied to assess the joint exposure effects of PM_2.5 chemical constituents on sleep. An interquartile range (IQR) increase in PM₁, PM_2.5, and PM₁₀ exposure during early life was associated with reduced nighttime sleep duration (β = −0.18; 95%CI: −0.32, −0.04; β = −0.23; 95%CI: −0.38, −0.08; β = −0.21; 95%CI: −0.35, −0.07). Combined early-life exposure to PM_2.5 constituents correlated with reduced nighttime sleep duration, with SO₄²⁻ being a major contributor. Regardless of particle size and PM_2.5 constituents, we identified consistent findings during pregnancy. We found positive associations between prenatal PM exposure and reduced nighttime sleep in two-year-old toddlers. The PM_2.5 constituent SO₄²⁻ appears to be the major contributor to these associations. Full article

This study investigates the interplay of green building and energy efficiency with innovative behavior and energy conservation measures within the economies of emerging Asia. It considers 14 countries for the period 2000–2023, built around the induced innovation hypothesis and Green Paradox. The empirical analysis deploys a layered econometric strategy comprising two-way fixed effects (FE-DK), MG-ARDL long-run estimation, Panel Smooth Transition Regression (PSTR), panel quantile regression, and Dumitrescu–Hurlin Granger causality tests. The findings are threefold. First, green building investment exerts a modest positive effect on eco-innovation but has no discernible effect on energy rebound at the aggregate level. Energy efficiency, however, is the primary rebound driver, with a near unit-elastic effect indicating that efficiency gains are almost entirely offset by compensating increases in aggregate energy demand. Second, non-linear analysis reveals a pronounced threshold effect: green building investment enhances eco-innovation at low levels but suppresses it beyond a critical threshold, a pattern corroborated by quantile regression across the performance distribution. Third, governance quality is the dominant institutional moderator of both outcomes, exerting particularly powerful leverage at high investment intensities. The Green Paradox is thus neither universal nor constant: its manifestation is contingent on investment scale and institutional quality. Full article

Real-time retail credit scoring is a data-intensive cognitive computing task. Each decision must fuse heterogeneous signals, execute a non-linear model, return a calibrated probability of default (PD), and emit a regulator-compliant local explanation within milliseconds. We address the most demanding segment of unsecured lending in Kazakhstan—Salary-Project-Independent (SPI) borrowers, whose principal income stream is not observable by the lender—and frame scoring as a constrained optimisation problem where we maximise discrimination subject to interpretability, latency, and calibration constraints. We propose a tenure-stratified hybrid framework that couples (i) an online weight-of-evidence logistic regression (WOE-LR) scorecard with (ii) an offline self-attention stacked ensemble (LightGBM, CatBoost, and a tabular self-attention network) whose calibrated PD is quantile-binned, WOE-encoded, and re-injected into the online scorecard as a single auditable predictor. On 551,962 production contracts that originated in 2022–2024, the repeat-client hybrid attains an area under the receiver operating characteristic curve (AUROC) of 0.826, a Gini coefficient of 0.65, and a Kolmogorov–Smirnov (KS) statistic of 0.495, preserving roughly half of the offline ensemble’s lift over the linear baseline (AUROC 0.79→0.897) while retaining a fully auditable twelve-coefficient scorecard in production. The new-client scorecard attains an AUROC of 0.741. Non-parametric isotonic recalibration reduces the expected calibration error from 0.27 to below 0.01 and raises the Hosmer–Lemeshow p-value above 0.99 without altering discrimination. The framework complies with the model risk standards of the Agency of the Republic of Kazakhstan for Regulation and Development of the Financial Market and is delivered as a Spark/MLOps reference architecture, illustrating how big data engineering, attention-based representation learning, and post hoc explanations can be co-designed for a high-stakes, high-throughput, regulated AI application. Full article

Chlorinated paraffins (CPs) are emerging contaminants with potential risks to the environment and human health, but their link between fine particulate matter-bound chlorinated paraffins (PM_2.5-bound CPs) and the risk of sleep disorders has not been reported. This large-scale, population-based study included 122,965 valid questionnaires from school-aged children in the Pearl River Delta region of China. Generalized linear mixed models, restricted cubic splines, weighted quantile sum regression and Quantile g-computation models were used to evaluate the individual and combined effects of PM_2.5-bound CPs on sleep disorders. Mediation analysis assessed the potential role of body weight. Our findings showed that an interquartile range (IQR) increase in PM_2.5-bound ∑CPs was linked to a higher risk of sleep disorders, and odds ratio ranging from 1.08 to 3.20. Short-chain chlorinated paraffins (SCCPs) were identified as key contributors of CPs and might interfere with important metabolic pathways. Further analysis indicated that body weight statistically accounted for part of the observed association between exposure to CPs and sleep disorders, with the proportion mediated ranging from 3.40% to 35.67%. Stratified analyses suggested stronger associations among girls and children under 12 years. These findings support prioritizing CPs management strategies to reduce their negative impact among children and adolescents. Full article

This study examines how village spatial institutional capacity (knowledge resources, relational resources, and mobilization capacity) and intra-organizational collaboration jointly shape the governance of agricultural irrigation systems. Using survey data from 840 households in six provinces of China’s Yellow River Basin, we employ OLS regression, bootstrapped quantile regression and moderation analysis. The empirical results indicate that both village spatial institutional capacity and internal collaboration significantly and positively affect comprehensive irrigation governance performance. Specifically, OLS results reveal that knowledge resources ($\mathsf{\beta }$ = 0.0029, p < 0.1), relational resources ($\mathsf{\beta }$ = 0.0711, p < 0.01), and mobilization capacity ($\mathsf{\beta }$ = 0.0236, p < 0.05) significantly enhance comprehensive performance, and internal collaboration exerts a significant positive moderating effect between institutional capacity and governance performance. Furthermore, quantile regression analysis reveals a non-linear distribution of this effect: the safeguarding role of institutional capacity is more prominent in the early stages of governance, whereas the driving role of internal collaboration shows an increasing trend in the middle and later stages. This study theoretically addresses the limitations of past reliance solely on the concept of “social capital” and provides robust empirical evidence for transitioning irrigation management from traditional top-down administrative dominance to a multi-centered, participatory self-governance model. Full article

The issue of water resources in a semi-arid country such as Morocco has been present for many years and is becoming increasingly critical. The droughts experienced over recent decades have demonstrated the country’s extreme vulnerability to any water deficit. In this context, the Berrechid plain represents a relevant case study illustrating both the practical and theoretical challenges of groundwater governance. The aquifer is heavily exploited to satisfy agricultural, industrial, and domestic needs. This study develops a hybrid “grey-box” modeling approach for predicting groundwater depth (GWD) fluctuations under climate change (CC). Unlike conventional black-box machine learning models, our framework combines a deterministic physical engine with a stochastic machine learning corrector. The physical component simulates aquifer mass balance using the Hargreaves method for evapotranspiration, linear drainage, climate memory via exponential decay, and an anthropogenic trend parameter (xi). The machine learning component—XGBoost with quantile regression—is trained exclusively on physical model residuals and predicts the 5th, 50th, and 95th percentiles, providing explicit 90% confidence intervals. Hydrological states (dry, normal, wet) are identified via K-means clustering for context-aware correction. The model is calibrated using historical data (1972–2019) and validated using blocked time-series cross-validation. Climate projections under the RCP 4.5 and RCP 8.5 scenarios were used to forecast GWD up to 2100. At piezometer 3933/20, the best performance was achieved, with an RMSE of 0.347 m and a KGE of 0.742 during the validation period. The proposed approach is suitable for seasonal GWD forecasting and offers practical value for water managers and decision-makers in the Berrechid region. Full article

Soil particle size fractions (PSFs)—sand, silt, and clay—are fundamental determinants of soil hydrological behavior, nutrient retention, and erodibility, yet their spatial prediction remains challenging due to the compositional nature of the data, unquantified prediction uncertainty, and limited interpretability of machine learning models. This study develops an integrated compositional mapping framework incorporating multi-source Sentinel-1/2 and topographic covariates, coupling the isometric log-ratio (ILR) transformation with Quantile Regression Forests (QRFs), a Monte Carlo simulation (MCS)-based latent-to-physical space uncertainty propagation strategy, and a Wrapper-SHAP attribution method to jointly address these challenges. The framework was evaluated across regional croplands in the central Shandong mountain-hilly region of China, using an elevation-stratified spatial cross-validation. Validations achieved R² values of 0.72, 0.61, and 0.59 for sand, silt, and clay, respectively, and a global Aitchison distance of 0.34. Critically, the MCS error propagation strategy effectively compensated for the probability distribution shift introduced by non-linear ILR back-transformation. This ensured that all predicted compositions strictly satisfied compositional closure and the [0, 100%] constraint, while aligning the prediction interval coverage probability (PICP) of each fraction closely with the 90% nominal level. Wrapper-SHAP overcame direct attribution limitations in compositional models, revealing the predictive associations of these multi-source covariates: high remote sensing-derived Bare Soil Index (BSI) and Moisture Stress Index (MSI) values primarily exhibited strong predictive associations with sand enrichment, whereas their lower values, combined with elevated Normalized Difference Moisture Index (NDMI), Enhanced Vegetation Index (EVI), and anthropogenic indicators, favored silt and clay accumulation. The proposed framework provides a transferable methodological reference for remote sensing-integrated compositional soil mapping with reliable uncertainty estimates and interpretable driver identification at regional scales. Full article

Urban vitality is regarded as a cornerstone of sustainable urban development. While land use mix (LUM) is widely acknowledged for fostering vitality, most empirical evidence relies on mean-effect models, neglecting the heterogeneous impacts across different vitality levels. This overlooks the complex, context-dependent nature of LUM and risks perpetuating one-size-fits-all planning. Based on a theoretical framework that links LUM analysis with contemporary urban revitalization, public governance, and smart city development discussions, this study leverages a Spatial Durbin Quantile Regression (SDQR) framework with multi-source geospatial data from 511 blocks in Ningbo, China, to systematically investigate the distributional heterogeneity of LUM’s effects on urban vitality. We decompose LUM into “diversity”, “proximity”, and “coordination” dimensions, revealing three distinct mechanisms across the vitality spectrum. Results show “coordination” acts as a fundamental “ignite” mechanism, consistently driving vitality across all quantiles, especially in new towns and low-vitality areas. “Diversity” primarily serves as a “boost” mechanism, enhancing vitality in medium-to-high vitality areas, demonstrating a non-linear, conditional effect. Crucially, “proximity” exhibits a novel “cap & siphon” mechanism: its direct effect is often insignificant or negative in low-vitality areas (suggesting structural mismatch), while its significant negative spatial spillover effect (siphon effect) across all quantiles, particularly in low-vitality zones, highlights intense inter-area competition. Furthermore, LUM’s direct effects tend to diminish in high-vitality areas, indicating a saturation or “cap” effect. By revealing these heterogeneous impacts and spatial spillover dynamics, this research refines the boundary conditions of classic mixed-use propositions and provides a differentiated planning paradigm, moving from universal zoning to context-specific, stage-calibrated interventions that address areas based on their current vitality levels, spatial interactions and governance contexts. Full article

Thickness accuracy is a critical quality indicator in medium and heavy plate production, as it directly affects material utilization, product performance, and manufacturing cost. The rolling process of medium and heavy plates is highly nonlinear. It also involves multivariable coupling and dynamic fluctuations in operating conditions. Therefore, achieving highly accurate and reliable thickness prediction in industrial applications remains a major challenge. To address this issue, this paper develops a joint point-interval prediction framework for medium and heavy plate thickness in industrial applications. First, recursive feature elimination with a LinearSVR estimator (LinearSVR-RFE) is employed to eliminate low-contribution features from the original process feature set, retain informative variables, and construct a compact and effective feature subset. Second, Bayesian optimization is employed to tune the hyperparameters of multiple machine learning regression models. A Stacking ensemble strategy is then adopted to improve the accuracy and robustness of point prediction under complex production conditions. Finally, quantile regression is introduced based on the optimal point prediction model to construct prediction intervals at multiple confidence levels. This provides uncertainty-aware results for production decision-making. Experimental results based on real industrial data from a 3500 mm medium and heavy plate production line show that the proposed framework achieves strong point prediction performance on the test set. The optimal Stacking model achieves a coefficient of determination (R²) of 0.9845 with a root mean square error (RMSE) of 0.73 mm on the test set. In addition, the framework produces prediction intervals with a good balance between coverage and compactness at confidence levels from 80% to 95%. For example, at the 90% confidence level, the interval prediction module achieves a PICP of 0.9043 and a PINAW of 0.0711. The results indicate that the proposed framework provides an effective solution for intelligent thickness prediction and quality evaluation in industrial rolling processes. It also shows good potential for engineering applications. Full article

This study enhances the knowledge base by providing an empirical inquiry into the asymmetric sensitivity of economic complexity (ECI) to changes in financial development (FD), using data from 30 African countries for the period of 1995–2023. To deliver robust estimates in the face of econometric pitfalls, this study employs estimators such as Hatemi-J data decomposition procedures, robust standard-error regression of Driscoll and Kraay, Feasible Generalised Least Squares, Lewbel’s IV-Two-Stage Least Squares, and Quantile regression via moments. The findings from the linear model indicate that FD enhances ECI upgrades in Africa. The findings provide robust evidence of asymmetric structures in ECI’s sensitivity to changes in FD. It highlights that both positive and negative change components (financial sector expansionary and contractionary policies, respectively) in the FD significantly contribute to ECI upgrades. These findings reveal the obscure aspects of how FD change components contribute differently to ECI upgrades in African countries. These findings highlight that expansionary financial sector policies aid the development of knowledge-based productivity, technology diffusion, and manufacturing capabilities, enabling the production of a chain of high-tech, high-quality, and globally competitive products for export. On the other hand, contractionary financial sector policies in African countries spur cumulative reductions in the channelling of financial resources and other technical support to ECI-impeding initiatives, thereby making more resources available to fund ECI-enhancing initiatives that aid the manufacturing of quality, competitive products for exports. This study draws and outlines relevant policy implications of the findings. Full article

Reliable climate data are essential for sustainable water management systems, especially under the challenges posed by climate change. In data-scarce regions, reanalysis products such as ERA5 can support flood and drought risk assessment and water security analysis. However, raw reanalysis precipitation is systematically biased relative to local observations and can distort hydrological indicators; bias correction is therefore needed. This study tests five bias correction techniques (Linear Scaling, Empirical Quantile Mapping, Quantile Mapping Spline Bias Correction, Mean Bias Subtraction, and Simple Linear Regression) on ERA5 precipitation data for Georgia, using classical and sliding window approaches at daily and monthly scales. Results show the importance of selecting the most appropriate method according to data availability and study objectives. The sliding window approach improved performance, especially at the daily scale, and distribution-based methods proved most effective in data-scarce regions. Full article

Quantifying the mixture effects on humans exposed remains challenging because mixture components are correlated and may act bidirectionally by exhibiting nonlinear dose-response relationships, which may contribute to subclinical organ dysfunction. The liver is a vital organ in the body with broad functions, making it vulnerable to injury as it is the first organ exposed to circulating toxicants, which can precipitate hepatic damage. Our study’s objective was to evaluate the combined and component-specific associations of a multi-chemical exposure mixture of heavy metals, polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxins (dioxins), and polychlorinated dibenzofurans (furans), with liver biomarkers, and to compare concentration-based results with the toxic equivalent (TEQ) potency of the weighted results for dioxin-like compounds. In an unweighted analytic sample of U.S. adults from NHANES 2003–2004 with 947 complete cases, we examined heavy metals (cadmium, lead, and mercury), PCBs (12 congeners), dioxins (7 congeners), and furans (10 congeners) in relation to eight liver biomarkers (albumin, ALP, ALT, AST, GGT, LDH, total bilirubin, and total protein). We applied multi-exposure linear regression, weighted quantile sum (WQS) regression, quantile g-computation (qgcomp), and Bayesian kernel machine regression (BKMR), with parallel TEQ-based models using WHO 2005 TEFs for dioxin-like PCBs, dioxins, and furans. Across mixture methods, the mixture structure was chemically sparse, with a limited set of recurring contributors. Total bilirubin showed the most consistent positive mixture association across qgcomp and BKMR and persisted under TEQ weighting, with prominent PCB- and dioxin-like contributions (notably PCB81/PCB TEQs and dioxin-related components). Albumin demonstrated inverse mixture patterns in BKMR and TEQ-BKMR, with dioxin-like components (notably Dioxin3 and Dioxin3_TEQ) repeatedly emerging as key drivers. For ALT, ALP, AST, GGT, LDH, and total protein, overall mixture effects were frequently attenuated or null in qgcomp despite structured component weights, indicating bidirectional sub-mixtures and internal counterbalancing. BKMR PIPs similarly concentrated on a small number of dominant predictors (e.g., lead for ALP, mercury for ALT, PCB28 for AST, and cadmium and PCB189 for LDH), while interaction summaries provided limited evidence of stable non-additivity. Using multiple complementary mixture methods, we identified outcome-specific mixture patterns suggesting hepatobiliary vulnerability. TEQ concordance supports toxicological relevance of the dioxin-like axis, while metals and non–dioxin-like mechanisms likely contribute additional pathways. Full article

Background: Dietary cystine may influence lipid metabolism, but epidemiological evidence in children is limited. This study aimed to investigate the association between dietary cystine intake and dyslipidemia and its subtypes in Chinese children. Methods: Data were derived from the China National Nutrition and Health Surveillance of Children and Lactating Mothers (CNNHSCLM). After propensity score matching (1:1, caliper = 0.2), 3676 children aged 6–17 years (1838 with dyslipidemia, 1838 controls) were included. The Quantile g-computation (qgcomp) model assessed the joint effect of 20 amino acids. Multivariate logistic regression, subgroup analysis, restricted cubic splines (RCS), and five machine learning models (including XGBoost with Shapley Additive Explanation (SHAP) analysis) were applied to evaluate the association between cystine intake and dyslipidemia. Results: The qgcomp model showed that cystine had a negative weighting contribution to reducing the risk of hypo-HDL cholesterolemia. Multivariate logistic regression revealed that cystine intake was significantly negatively correlated with hypo-HDL cholesterolemia (OR = 0.67, 95%CI: 0.53–0.86, p = 0.002) and total dyslipidemia (OR = 0.84, 95%CI: 0.74–0.96, p = 0.010), but not with other subtypes. Subgroup analyses indicated interactions with BMI and sex. RCS showed a non-linear dose–response relationship for hypo-HDL cholesterolemia and a linear negative relationship for total dyslipidemia. The XGBoost model achieved the best predictive performance (AUC = 0.902), and SHAP analysis identified cystine as the most important feature inversely associated with dyslipidemia. Decision curve analysis confirmed its clinical net benefit. Conclusions: Dietary cystine intake is negatively associated with the risk of hypo-HDL cholesterolemia and total dyslipidemia in children, and cystine is an important negative correlate of dyslipidemia. These findings provide new scientific evidence for dietary prevention of dyslipidemia in children. Full article

This research article explores whether the climate transition risk (CTR) and climate physical risk (CPR) transmit greater shocks towards the sustainable, gold-backed, energy-related and Sharia-compliant cryptocurrencies during bullish market conditions as compared with the normal and bearish market conditions. We employ the novel quantile vector auto-regression (QVAR)-based connectivity framework. Overall findings suggested that CPR and CTR transmitted greater shocks towards cryptocurrency classes during extremely high and lower quantiles as compared with the median quantile. This U-shaped and non-linear climate risks shock transmission indicates that Sharia-compliant, energy-related and gold-backed cryptocurrencies become more vulnerable during extreme market conditions (higher and lower quantiles) and may not consistently serve as reliable hedging or diversification instruments, particularly during periods of heightened climate uncertainty. Overall findings suggested that both the CPR and CTR transmitted greater shocks towards energy-related, gold-backed, and Sharia-compliant cryptocurrencies as compared with the sustainable cryptocurrencies, across all the quantiles. Therefore, sustainable cryptocurrencies, particularly those with energy-efficient consensus mechanisms such as Stellar, Cardano and Ripple, exhibited resilience to climate risks and can therefore function as stabilizing core holdings in diversified portfolios. Fund managers should incorporate a rebalancing strategy that increases allocation to these climate-resilient, sustainable digital assets during periods of elevated climate risk. Fund managers should integrate CPR and CTR into the quantile-domain forecasting frameworks for predicting digital asset market returns to enhance financial stability. Portfolio managers should undertake dynamic and quantile-contingent climate risk hedging strategies that account for tail-risk exposure rather than relying on average market behavior. Full article