Search

This study constructs a measure of audit quality that captures the effect of potential factors that are generally unobservable to people outside of the audit firm or client company. Using machine learning and a wide range of data describing audit firm characteristics, audit partners, and public companies in China, this paper constructs the "surprise score," a new measure of audit quality, calculated as the difference between the predicted probability and the actual value of an audit quality-related event (i.e., the existence of material misstatements, audit adjustments, and nonclean audit opinions). The effectiveness of the surprise score is validated by testing the association between the surprise score and penalties or audit firm changes. The proposed approach is applied to U.S. data to generalize its application. The surprise score adds value to existing audit quality measures and can help regulators to make better-informed decisions about audit quality. Data Availability: Except for the data privately provided by CICPA and MFC, other datasets are available from the public sources cited in the text. JEL Classifications: M41; M42. [ABSTRACT FROM AUTHOR]

Nonlinearly constrained nonconvex and nonsmooth optimization models play an increasingly important role in machine learning, statistics, and data analytics. In this paper, based on the augmented Lagrangian function, we introduce a flexible first-order primal-dual method, to be called nonconvex auxiliary problem principle of augmented Lagrangian (NAPP-AL), for solving a class of nonlinearly constrained nonconvex and nonsmooth optimization problems. We demonstrate that NAPP-AL converges to a stationary solution at the rate of o(1/k) , where k is the number of iterations. Moreover, under an additional error bound condition (to be called HVP-EB in the paper) with exponent θ∈(0,1) , we further show the global convergence of NAPP-AL. Additionally, if θ∈(0,12] , then we furthermore show that the convergence rate is in fact linear. Finally, we show that the well-known Kurdyka-Łojasiewicz property and the Hölderian metric subregularity imply the aforementioned HVP-EB condition. We demonstrate that under mild conditions, NAPP-AL can also be interpreted as a variant of the forward-backward operator splitting method in this context. Funding: This work was supported by the National Natural Science Foundation of China [Grant 71871140]. [ABSTRACT FROM AUTHOR]

Background: Plasmodium falciparum cases are rising in China due to the imported malaria cases from African countries. The main goal of this study is to examine the impact of imported malaria cases in African countries on the rise of P. falciparum cases in China before and during the COVID-19 pandemic. Methods: A generalized regression model was used to investigate the association of time trends between imported malaria cases from 45 African countries and P. falciparum cases in 31 provinces of China from 2012 to 2018 before the COVID-19 pandemic and during the COVID-19 pandemic from October 2020 to May 2021. Based on the analysis, we proposed a statistical and deep learning hybrid approach to model the resurgence of malaria in China using monthly data of P. falciparum from 2004 to 2016. This study builds a hybrid model known as the ARIMA-GRU approach for modeling the P. falciparum cases in all provinces of China and the number of malaria deaths in China before and during the COVID-19 pandemic. Results: The analysis showed an emerging link between the rise of imported malaria cases from Africa and P. falciparum cases in many provinces of China. Many imported malaria cases from Africa were P. falciparum cases. The proposed deep learning model achieved a high prediction accuracy score on the testing dataset of 96%. Conclusion: The study provided an analysis of the reduction of P. falciparum cases and deaths caused by imported P. falciparum cases during the COVID-19 pandemic due to the control measures regarding the limitation of international travel in China. The Chinese government has to prepare the imported malaria control measures after the normalization of international travel, to prevent the resurgence of malaria disease in China. [ABSTRACT FROM AUTHOR]

The lack of information on the extinction risk of most species is a fundamental challenge in prioritizing conservation strategies and bending the curve of current biodiversity decline. Machine learning methods have shown promising potential to fill this gap, but their applicability remains to be validated at different taxa (especially aquatic species) and spatial scales. We assessed the extinction risk of 1162 freshwater fish species in China that have not yet been included in the latest IUCN Red List using multiple neural network algorithms based on datasets of species occurrences, biological traits, phylogeny, and relevant environmental layers. The best deep learning models dramatically improved the assessment coverage from 29.9% (496 species) to 93.2–93.9% (1545–1557 species) of the whole fauna with an accuracy of 95.4–99.0%. By combining our prediction results with the IUCN Red List, we found that 23.8–26.5% (394–440 species) of Chinese freshwater fishes were identified as possibly threatened species, which is roughly four times the IUCN assessment. Newly assessed species and threatened species were mainly from the orders Cypriniformes (prediction added: 837–846 species; final threatened: 325–350 species), Siluriformes (113–122; 28–37) and Perciformes (74–76; 18–25). The increase in threatened species richness based on predictions was led by the upper reaches of the Pearl and Yangtze. Overall, our findings suggest that deep learning algorithms can provide robust and time-saving assessments of extinction risk for entire freshwater fish fauna on a large national scale, thereby facilitating relevant conservation prioritization. [ABSTRACT FROM AUTHOR]

The escalating levels of carbon dioxide (CO2) emissions represent the primary driver of global warming, and addressing them is of paramount importance. Timely and accurate prediction, as well as effective control of CO2 emissions, are pivotal for guiding mitigation measures. This paper aims to select the best prediction model for near-real-time daily CO2 emissions in China. The prediction models are based on univariate daily time-series data spanning January 1st, 2020, to September 30st, 2022. Six models are proposed, including three statistical models: grey prediction (GM(1,1)), autoregressive integrated moving average (ARIMA), and seasonal autoregressive integrated moving average with exogenous factors (SARIMAX), and three machine learning models: artificial neural network (ANN), random forest (RF), and long short-term memory (LSTM). The performance of these six models is evaluated using five criteria: mean squared error (MSE), root-mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and coefficient of determination (R2). Our findings reveal that the three machine learning models consistently outperform the three statistical models across all five criteria. Among them, the LSTM model demonstrates exceptional performance for daily CO2 emission prediction, boasting an impressively low MSE value of 3.5179e−04, an RMSE value of 0.0187, an MAE value of 0.0140, an MAPE value of 14.8291%, and a high R2 value of 0.9844. This underscores the robustness of the LSTM model in capturing and predicting complex emission patterns, positioning it as the most suitable option for near-real-time daily CO2 emission prediction based on the provided daily time series data. Moreover, our study's results provide valuable insights into emissions forecasting, enabling data-driven decision-making for policymakers and stakeholders. The accurate and timely predictions offered by the LSTM model can aid in the formulation of effective strategies to mitigate carbon emissions, contributing to a more sustainable future. Furthermore, the findings of this study can enhance our understanding of the dynamics of CO2 emissions, leading to more informed environmental policies and actions aimed at reducing carbon emissions. [ABSTRACT FROM AUTHOR]

Solar and wind resources are vital for the sustainable energy transition. Although renewable potentials have been widely assessed in existing literature, few studies have examined the statistical characteristics of the inherent renewable uncertainties arising from natural randomness, which is inevitable in stochastic-aware research and applications. Here we develop a rule-of-thumb statistical learning model for wind and solar power prediction and generate a year-long dataset of hourly prediction errors of 30 provinces in China. We reveal diversified spatiotemporal distribution patterns of prediction errors, indicating that over 60% of wind prediction errors and 50% of solar prediction errors arise from scenarios with high utilization rates. The first-order difference and peak ratio of generation series are two primary indicators explaining the uncertainty distribution. Additionally, we analyze the seasonal distributions of the provincial prediction errors that reveal a consistent law in China. Finally, policies including incentive improvements and interprovincial scheduling are suggested. Renewable uncertainty analysis is vital for stochastic-aware research. This study generates a benchmark dataset of year-long hourly renewable prediction errors in China, and reveals the law of the spatiotemporal distribution of renewable uncertainty. [ABSTRACT FROM AUTHOR]

Archaeological predictive modeling (APM) is an essential method for quantitatively assessing the probability of archaeological sites present in a region. It is a necessary tool for archaeological research and cultural heritage management. In particular, the predictive modeling process could help us understand the relationship between past human civilizations and the natural environment; moreover, a better understanding of the mechanisms of the human–land relationship can provide new ideas for sustainable development. This study aims to investigate the impact of topographic and hydrological factors on archaeological sites in the Japanese archipelago and Shaanxi Province, China and proposes a hybrid integration approach for APM. This approach employed a conditional attention mechanism (AM) using deep learning and a frequency ratio (FR) model, in addition to a separate FR model and the widely-used machine learning MaxEnt method. The models' outcomes were cross-checked using the four-fold cross-validation method, and the models' performances were compared using the area under the receiver operating characteristic curve (AUC) and Kvamme's Gain. The results showed that in both study areas, the AM_FR model exhibited the most satisfactory performances. [ABSTRACT FROM AUTHOR]

Purpose: Although soil total phosphorus (TP) is a primary and essential large element reflecting the soil fertility in agricultural ecosystems, studies on model development of TP and its differences between wheat and paddy lands after a long cultivation history at a regional scale are still limited. Hence, a comparison model of TP with different learning methods and datasets were built, and the relationship between environmental factors and TP were discussed. Methods: TP from a long cultivation of either wheat or paddy agriculture systems was investigated, and the regression between TP and climate parameters (air temperatures, precipitation, humidity, and atmospheric pressure) and latitude were analyzed. A comparison of model development with six learning methods, including one statistical learning method (linear) and five machine learning methods (support vector regression, decision tree, random forest, XGBoost, and LightGBM), and two datasets (0–20 and 0–170-cm soil layers) was made. The models were evaluated by the root mean squared error (RMSE), mean deviation (RMD), mean absolute error (MAE), and model effective (EF). Results: The results showed that the TP content of the top soil layer in wheat lands (0.89 ± 0.01 g kg−1) was significantly higher than that of paddy lands (0.63 ± 0.01 g kg−1). The annual average precipitation, humidity, and air temperature had significant negative relationships with TP content, while the annual average atmospheric pressure and latitude had significant positive relationships with TP. Most machine learning methods showed better performances than that of a statistical learning method with the highest r2 of 0.82. The different datasets used for model development had no significant effect on model performances. Conclusion: The average TP content of the top soil layer tends to be greater in wheat lands than that in paddy lands after a long cultivation. Other than the different statistical parameters (the average, maximum, and minimum values) of each climate parameter, comprehensive climate parameters including the annual, semiannual, quarterly, and monthly air temperature, precipitation, humidity, and atmospheric pressure should be considered for further model development. Although different datasets in variable soil depth had no significant effect on model performances, machine learning methods such as random forest, XGBoost, and LightGBM are recommended for better performance than a linear learning method for soil TP model development. It is recommended that a comparison of different machine learning methods will help build a stronger model in similar studies. [ABSTRACT FROM AUTHOR]

Ensemble learning methods, such as boosting, focus on producing a strong classifier based on numerous weak classifiers. In this paper, we develop a novel ensemble learning method called rescaled boosting with truncation (ReBooT) for binary classification by combining well-known rescaling and regularization ideas in boosting. Theoretically, we present some sufficient conditions for the convergence of ReBooT, derive an almost optimal numerical convergence rate, and deduce fast-learning rates in the framework of statistical learning theory. Experimentally, we conduct both toy simulations and four real-world data runs to show the power of ReBooT. Our results show that, compared with the existing boosting algorithms, ReBooT possesses better learning performance and interpretability in terms of solid theoretical guarantees, perfect structure constraints, and good prediction performance. History: Accepted by Ram Ramesh, Area Editor for Data Science & Machine Learning Funding: This work was supported by the National Natural Science Foundation of China [Grants 11971374, 61772374, and 61876133]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/ijoc.2022.1224. [ABSTRACT FROM AUTHOR]

Debris flows, triggered by dual interferences extrinsically and intrinsically, have been widespread in China. The debris-flow susceptibility (DFS) assessment is acknowledged as the benchmark for the mitigation and prevention of debris flow risks, but DFS assessments at the national level are lacking. The role of human activities in the DFS assessment has always been overlooked. On the basis of a detailed inventory of debris-flow sites and a large set of environmental and human-related characteristics, this research presents the comparative performance of the well-known information value (IV), logistic regression (LR) and random forest (RF) models for DFS assessments in China. Twelve causative factors, namely, elevation, slope, aspect, rainfall, the normalized difference vegetation index (NDVI), land use, landform, geology, distance to faults, density of villages, distance to rivers and distance to roads, were considered. Debris-flow susceptibility maps were then generated after the nonlinear relationship between the debris-flow occurrence and the causative factors was captured. Finally, the predictive performance of the three maps was evaluated through receiver operating characteristic (ROC) curves, and the validation results showed that areas under the ROC curves were 81.98%, 79.96% and 97.38% for the IV, LR and RF models, respectively, indicating that the RF model outperformed the other two traditional statistical methods. The importance ranking of the RF model also revealed that distance to roads, slope and rainfall dominated the spatial distribution of debris flows. This is the first experiment to compare between the traditional statistical and machine learning methods in DFS studies for the whole of China. Our results could provide some empirical support for China's policymakers and local practitioners in their efforts to enable residents to be less vulnerable to disasters. [ABSTRACT FROM AUTHOR]

Evaluating the influence of anthropogenic-emission changes on air quality requires accounting for the influence of meteorological variability. Statistical methods such as multiple linear regression (MLR) models with basic meteorological variables are often used to remove meteorological variability and estimate trends in measured pollutant concentrations attributable to emission changes. However, the ability of these widely used statistical approaches to correct for meteorological variability remains unknown, limiting their usefulness in the real-world policy evaluations. Here, we quantify the performance of MLR and other quantitative methods using simulations from a chemical transport model, GEOS-Chem, as a synthetic dataset. Focusing on the impacts of anthropogenic-emission changes in the US (2011 to 2017) and China (2013 to 2017) on PM 2.5 and O3 , we show that widely used regression methods do not perform well in correcting for meteorological variability and identifying long-term trends in ambient pollution related to changes in emissions. The estimation errors, characterized as the differences between meteorology-corrected trends and emission-driven trends under constant meteorology scenarios, can be reduced by 30 %–42 % using a random forest model that incorporates both local- and regional-scale meteorological features. We further design a correction method based on GEOS-Chem simulations with constant-emission input and quantify the degree to which anthropogenic emissions and meteorological influences are inseparable, due to their process-based interactions. We conclude by providing recommendations for evaluating the impacts of anthropogenic-emission changes on air quality using statistical approaches. [ABSTRACT FROM AUTHOR]

A key element for reducing energy consumption and improving thermal comfort on high-speed rail is controlling air-conditioning temperature. Accurate prediction of air supply temperature is aimed at improving control effects. Existing studies of supply air temperature prediction models are interdisciplinary, involving heat transfer science and computer science, where the problem is defined as time-series prediction. However, the model is widely accepted as a complex model that is nonlinear and dynamic. That makes it difficult for existing statistical and deep learning methods, e.g., autoregressive integrated moving average model (ARIMA), convolutional neural network (CNN), and long short-term memory network (LSTM), to fully capture the interaction between these variables and provide accurate prediction results. Recent studies have shown the potential of the Transformer to increase the prediction capacity. This paper offers an improved temporal fusion transformers (TFT) prediction model for supply air temperature in high-speed train carriages to tackle these challenges, with two improvements: (i) Double-convolutional residual encoder structure based on dilated causal convolution; (ii) Spatio-temporal double-gated structure based on Gated Linear Units. Moreover, this study designs a loss function suitable for general long sequence time-series forecast tasks for temperature forecasting. Empirical simulations using a high-speed rail air-conditioning operation dataset at a specific location in China show that the temperature prediction of the two units using the improved TFT model improves the MAPE by 21.70% and 11.73%, respectively the original model. Furthermore, experiments demonstrate that the model effectively outperforms seven popular methods on time series computing tasks, and the attention of the prediction problem in the time dimension is analyzed. [ABSTRACT FROM AUTHOR]

Reliable and fine-resolution electric power consumption (EPC) is essential for effective urban electricity allocation and planning. Currently, EPC data exists mainly as statistics with low resolution. Many studies estimate fine-resolution EPC based on the positive correction between stable nighttime light and EPC distribution. However, EPC is related to various factors other than nighttime light and is spatially non-stationary. Yet this has been ignored in current research. This study developed a novel method to estimate EPC at 500 m resolution by considering spatially non-stationary through fusing geospatial data and high-resolution satellite images. Deep transfer learning and statistical methods were used to extract socio-economic, population density, and landscape features to describe EPC distribution from multi-source geospatial data. Finally, a random forest regression (RFR) model with features and EPC statistics is established to estimate fine-resolution EPC. A study area of Shenzhen city, China, is employed to evaluate the proposed method. The R 2 between predicted EPC and statistical EPC is 0.82 at sub-district level in 2013, which is higher than an existing EPC product (Shi's product) with R 2 = 0.46 , illustrating the effectiveness of the proposed method. Moreover, the EPC distribution for Shenzhen from 2013 to 2019 was estimated. Furthermore, the spatiotemporal dynamic of EPC was analyzed at the pixel and sub-district levels. [ABSTRACT FROM AUTHOR]

As one of the most widely used languages in the world, English plays a vital role in the communication between China and the world. However, grammar learning in English is a difficult and long process for English learners. Especially in English writing, English learners will inevitably make various grammatical writing errors. Therefore, it is extremely important to develop a model for correcting various writing errors in English writing. This can not only be used for automatic inspection and proofreading of English texts but also enable students to achieve the purpose of autonomous practice. This paper constructs an English writing error correction model and applies it to the actual system to realize automatic checking and correction of writing errors in English composition. This paper uses the deep learning model of Seq2Seq_Attention model and transformer model to eliminate deep-level errors. Statistical learning is combined with deep learning and adopted a model integration method. The output of each model is sent to the n-gram language model for scoring, and the highest score is selected as output. [ABSTRACT FROM AUTHOR]

We propose a dust‐event detection and tracking procedure based on air quality data from the ground monitoring network by detecting temporal and spatial change‐points in PM10 concentration. It supplements the existing remote sensing based approach with high temporal resolution and better weather adaptivity. Applications of the procedure on the labeled data showed its having high discriminating power for dust events, pollution events, and clean periods. Our study finds changing correlation patterns between PM10 and other air pollutants at the start of the dust events, which are utilized to enhance the discriminating power of the dust‐event detection procedure. The detection and tracking procedure allows the construction of transport networks of the dust‐events as well as the identification of the source regions and the transportation pattern, and assess the intensity and severity of the dust‐events in North China. Our analysis find the dust‐events contributed to 23.3%–34.6% for PM10 and 18.2–33.2% for PM2.5 in the source regions and 2.0%–7.3% and 0.8%–4.0%, respectively, in the down‐stream provinces in the spring season from 2015 to 2020. Plain Language Summary: Dust storm can be tracked via the air quality monitoring network via a statistical semi‐supervised learning algorithm. Key Points: Develop a dust event detection procedure based on air quality data with finer spatial‐temporal resolution and better weather adaptationConstruct dust transport networks and find altering dependence between PM10 and CO, SO2 and NO2 at the start of dust eventsDust‐events contributed to 23.3%–34.6% for PM10 and 18.2–33.2% for PM2.5 in the source regions and 2.0%–7.3% and 0.8%–4.0%, respectively, in the down‐stream provinces in spring in North China [ABSTRACT FROM AUTHOR]

Background. From Ebola, Zika, to the latest COVID-19 pandemic, outbreaks of highly infectious diseases continue to reveal severe consequences of social and health inequalities. People from low socioeconomic and educational backgrounds as well as low health literacy tend to be affected by the uncertainty, complexity, volatility, and progressiveness of public health crises and emergencies. A key lesson that governments have taken from the ongoing coronavirus pandemic is the importance of developing and disseminating highly accessible, actionable, inclusive, coherent public health advice, which represent a critical tool to help people with diverse cultural, educational backgrounds and varying abilities to effectively implement health policies at the grassroots level. Objective. We aimed to translate the best practices of accessible, inclusive public health advice (purposefully designed for people with low socioeconomic and educational background, health literacy levels, limited English proficiency, and cognitive/functional impairments) on COVID-19 from health authorities in English-speaking multicultural countries (USA, Australia, and UK) to adaptive tools for the evaluation of the accessibility of public health advice in other languages. Methods. We developed an optimised Bayesian classifier to produce probabilistic prediction of the accessibility of official health advice among vulnerable people including migrants and foreigners living in China. We developed an adaptive statistical formula for the rapid evaluation of the accessibility of health advice among vulnerable people in China. Results. Our study provides needed research tools to fill in a persistent gap in Chinese public health research on accessible, inclusive communication of infectious diseases' prevention and management. For the probabilistic prediction, using the optimised Bayesian machine learning classifier (GNB), the largest positive likelihood ratio (LR+) 16.685 (95% confidence interval: 4.35, 64.04) was identified when the probability threshold was set at 0.2 (sensitivity: 0.98; specificity: 0.94). Conclusion. Effective communication of health risks through accessible, inclusive, actionable public advice represents a powerful tool to reduce health inequalities amidst health crises and emergencies. Our study translated the best-practice public health advice developed during the pandemic into intuitive machine learning classifiers for health authorities to develop evidence-based guidelines of accessible health advice. In addition, we developed adaptive statistical tools for frontline health professionals to assess accessibility of public health advice for people from non-English speaking backgrounds. [ABSTRACT FROM AUTHOR]

Addressing the urgent need for accurate carbon emissions forecasting to support global emissions reduction goals, this paper introduces a novel approach for carbon emissions prediction that dynamically considers the spatial–temporal correlation of carbon emissions across forecasting targets. Traditional statistical and machine learning models have limitations, such as oversimplification and inefficiency in capturing evolving dynamics among multiple forecasting targets, alongside a focus on longer forecasting horizons like monthly or yearly. To tackle these, we propose an innovative Dynamic Spatial–Temporal Graph Convolutional Recurrent Network (DSTGCRN), a blend of graph convolutional and recurrent neural network structures customized for up-to-date multi-regional spatial–temporal predictions at a daily level. Comprehensive empirical analyses on datasets from China, the US, and the EU confirm DSTGCRN's superior performance and robustness across diverse geographical contexts. It outperforms the second-best among 10 baseline models, achieving 40.6%, 24.6%, and 38.5% improvements in MAE, RMSE, and MAPE, respectively, across various horizons. The incorporation of environmental data, including temperature and Air Quality Index, as supplementary predictors in the DSTGCRN model, has demonstrated its efficacy in the context of the China dataset. The DSTGCRN model's improved accuracy and daily multi-regional forecasting deepen emission dynamics understanding, supporting the development of informed and region-specific environmental policies that can respond promptly to changes. [ABSTRACT FROM AUTHOR]

Predicting water demand helps decision-makers allocate regional water resources efficiently, thereby preventing water waste and shortage. The aim of this study is to predict water demand in the Beijing–Tianjin–Hebei region of North China. The explanatory variables associated with economy, community, water use, and resource availability were identified. Eleven statistical and machine learning models were built, which used data covering the 2004–2019 period. Interpolation and extrapolation scenarios were conducted to find the most suitable predictive model. The results suggest that the gradient boosting decision tree (GBDT) model demonstrates the best prediction performance in the two scenarios. The model was further tested for three other regions in China, and its robustness was validated. The water demand in 2020–2021 was provided. The results show that the identified explanatory variables were effective in water demand prediction. The machine learning models outperformed the statistical models, with the ensemble models being superior to the single predictor models. The best predictive model can also be applied to other regions to help forecast water demand to ensure sustainable water resource management. [ABSTRACT FROM AUTHOR]

Purpose: Understanding the microbial linkages among the soils, plants, and animals is crucial for maintaining the balance of the grazed grassland ecosystem. However, previous studies always focused on the biotopes of soil, phyllosphere, and feces separately and little has been known about microbial distributions and migrations among these biotopes. In this study, systematic surveys about the overlap and differentiation among various microbiotas of biotopes and how the environmental filter on microorganisms served for the ecosystem were conducted. Materials and methods: Soil, phyllosphere, and feces samples were collected from grazed and ungrazed grassland in Inner Mongolia, China. High-throughput sequencing and qPCR were employed to obtain the genome data. The results were analyzed by various statistical tools such as genomic analysis and machine learning classification to describe the disparities and linkages among the microbiotas of three biotopes. Results and discussion: Our findings revealed that the biotopes drove the distinct microbial community assemblages with various richness, β-diversity and composition. The substantial overlaps between soil and phyllosphere in fungi, bacteria and archaea indicated that soil played the role of the microbial source for the phyllosphere. Communal OTUs could be found between phyllosphere and feces. Nevertheless, Ascomycota were the only microorganisms migrating among all the biotopes massively. After the long-term impact of feces via grazing, the soil and phyllosphere microbiota also altered significantly. Conclusions: Biotopes drive the discrepancy of microbiota distribution among the soil, phyllosphere, and feces. Soil could potentially perform as the microbial reservoir for the phyllosphere. The phyllosphere plays the role of bridge to link soil and feces biotope through its fluctuated condition. However, there was only fungal migration running through the ecosystem to link all the biotopes. These findings promoted our understanding of the biotope contribution to microbial migration and improved the knowledge of microbial linkages in the grazed grassland ecosystem. [ABSTRACT FROM AUTHOR]

This paper uses statistical learning theory and big data analysis to study the energy consumption structure of China from qualitative and quantitative aspects during COVID-19. According to the domestic and foreign scholars' research on the optimization of energy consumption structure, the carbon emission factor is considered in the optimization of energy consumption structure. Taking the minimum energy consumption cost and carbon dioxide emission as the objective function, the carbon dioxide emission is taken as the objective function, and the total energy consumption and various energy consumption proportions as the constraint conditions, the multi-objective planning method is used to evaluate the energy consumption structure of China. The optimization model of source consumption structure is analyzed, and the medium and long-term energy transformation path and optimization model under low-carbon constraints are studied. Combined with the experimental algorithms related to big data, it is concluded that China's economic development mainly depends on a large amount of energy consumption during the COVID-19 period. On this basis, some suggestions are put forward to realize the sustainable development of China's economy and energy. [ABSTRACT FROM AUTHOR]

Although numerous studies have reported the influencing factors of polycyclic aromatic hydrocarbons (PAHs) in surface soil from source, process or soil perspectives, the mechanism of PAHs heterogeneity in surface soil are still not well understood. In this study, the effects of 16 PAHs in surface soil of China sampled between 2003 and 2020 with their 17 "source-process-sink" factors at 1 km resolution (N = 660)) were explored using deep learning (eXtreme Gradient Boosting) to mine key information from complex dataset under the optimized parameters (i.e., learning rate = 0.05, maximum depth = 5, sub-sample = 0.8). It was observed that top five factors of 16 PAH had the largest cumulative contribution (i.e., from 84.8% to 98.1%) on their soil concentrations. PAH emission was the predominant driver, and its effect on soil PAH increases with increasing log K ow. Soil was the second driver, in which clay can promote the partition of PAHs with low or middle log K ow. However, sand can accumulate those congeners with high log K ow. Moreover, the deep learning plus geo-statistical models (with low deviation for testing dataset (N = 283)) were capable of predicting soil PAH concentrations using their drivers with high accuracy. This study improved the understanding of the environmental fate and spatial variability of soil PAHs, as well as provided a novel technique (i.e., deep learning coupled with geo-statistics) for accurate prediction of soil pollutants. [Display omitted] • Contribution of "source-process-sink" factors to soil PAHs were quantified. • Five key drivers of each soil PAH were distinguished. • Contributions of drivers to soil PAHs were ranged from 84.8% to 98.1%. • More attention should be paid to PAHs pollution in soil along "Hu line". • A hybrid model was proposed to accurately model surface soil PAHs. [ABSTRACT FROM AUTHOR]

In recent years, peer-to-peer (P2P) lending in China, which is a new form of unsecured financing that uses the Internet, has boomed, but the consequent credit risk problems are inevitable. A key challenge facing P2P lending platforms is accurately predicting the default probability of the borrower of each loan using the default prediction model, which effectively helps the P2P lending platform avoid credit risks. The traditional default prediction model based on machine learning and statistical learning does not meet the needs of P2P lending platforms in terms of default risk prediction because for data-driven P2P lending, credit data have a large number of missing values, are high-dimensional and have class-imbalanced problems, which makes it difficult to effectively train the default risk prediction model. To solve the above problems, this paper proposes a new default risk prediction model based on heterogeneous ensemble learning. Three individual classifiers, extreme gradient boosting (XGBoost), a deep neural network (DNN) and logistic regression (LR), are used simultaneously with a liner weight ensemble strategy. In particular, this model is able to process missing values. After generating discrete and rank features, this model adds missing values to the model for self-training. Then, the hyperparameters are optimized by the XGBoost model to improve the performance of the prediction model. Finally, compared with the benchmark model, the proposed method significantly improves the accuracy of the prediction results. In conclusion, the prediction method proposed in this paper solves the class-imbalanced problem. [ABSTRACT FROM AUTHOR]

Digital soil mapping (DSM) often relies on existing soil samples obtained from various sources. However, the spatial distribution of such soil samples can be biased, for example, towards areas of better accessibility. Such biased coverage over the geographic space (i.e., spatial bias) often leads to biased coverage of the soil samples over the environmental covariate space. As a result, spatial bias degrades the correlation or statistical relationship between samples and covariates in the study area and impedes DSM accuracy. This paper presents a representativeness heuristic for mitigating spatial bias in existing soil samples for improving DSM accuracy. The key idea of the heuristic was to define and quantify sample representativeness as the goodness-of-coverage of the soil samples over the environmental covariate space. Spatial bias was then mitigated by weighting the samples towards maximizing their representativeness. Determination of the sample weights was conceived as an optimization problem and accordingly the optimal weights were determined using a genetic algorithm. To evaluate the effectiveness of the representativeness heuristic, a case study of mapping soil organic matter (SOM) content using existing soil samples was conducted in Heshan study area, northeastern China. Results showed that weighting soil samples using the optimal weights determined from the representativeness heuristic improved SOM content mapping accuracy. Moreover, a positive relationship between sample representativeness and mapping accuracy was observed, suggesting sample representativeness is an effective indicator of mapping accuracy. Additionally, the determined optimal weights were informative of individual sample importance and thus can be used as guidance to filter existing soil samples to improve DSM accuracy. • A novel approach for mitigating spatial bias in samples for digital soil mapping • The approach was based on quantification of the representativeness of soil samples. • Bias was mitigated by weighting samples to maximize their representativeness. • As demonstrated in a case study, the approach can effectively improve DSM accuracy. • It is applicable to many statistical and machine learning approaches for DSM. [ABSTRACT FROM AUTHOR]

Landslide susceptibility mapping is vital for landslide risk management and urban planning. In this study, we used three statistical models [frequency ratio, certainty factor and index of entropy (IOE)] and a machine learning model [random forest (RF)] for landslide susceptibility mapping in Wanzhou County, China. First, a landslide inventory map was prepared using earlier geotechnical investigation reports, aerial images, and field surveys. Then, the redundant factors were excluded from the initial fourteen landslide causal factors via factor correlation analysis. To determine the most effective causal factors, landslide susceptibility evaluations were performed based on four cases with different combinations of factors ("cases"). In the analysis, 465 (70%) landslide locations were randomly selected for model training, and 200 (30%) landslide locations were selected for verification. The results showed that case 3 produced the best performance for the statistical models and that case 2 produced the best performance for the RF model. Finally, the receiver operating characteristic (ROC) curve was used to verify the accuracy of each model's results for its respective optimal case. The ROC curve analysis showed that the machine learning model performed better than the other three models, and among the three statistical models, the IOE model with weight coefficients was superior. [ABSTRACT FROM AUTHOR]

The engineering classification of rock masses is the basis of rock engineering design and construction. We propose and apply a quick basic quality (BQ) classification method based on the standard BQ method of China to classify the quality grade of the rock mass around tunnels along the Ningguo-Huangshan Expressway during the construction period. Moreover, the joint continuity and surface roughness of the controlled key joint are added to the classification indices of the quick BQ method to address shortcomings of the standard BQ classification method. Therefore, an improved BQ classification method for rock mass is proposed. According to the BQ method, different personnel might select different values of correction coefficient that result in divergences in the result of rock mass classification. In order to solve this problem, the Genetic algorithm (GA) and support vector classification (SVC) coupling algorithm is introduced into the field of engineering rock mass classification. GA is used to automatically search for the optimal SVC parameters during the training process of samples. By training the classification samples of rock mass around a tunnel using the improved BQ method during the tunnel construction period, an intelligent SVC classification model is constructed with inputs based on eight classification indices and an output of the BQ quality grade. To verify the reliability and accuracy of the model, the SVC model is used to evaluate the quality grade of the rock mass around tunnel in other cross sections of the tunnels along the Ningguo-Huangshan Expressway. Only one section classification result differed from those of the improved BQ method in a total of 20 sections. In contrast, three section classification results based on the BP neural network (BPNN) model were inconsistent with those of the improved BQ method. Therefore, the proposed SVC model displays a higher rate of correct classification relative to that of the BPNN model. Meanwhile, the use of this SVC model can avoid the divergence among different people on the classification result of rock mass around a tunnel, which provides an effective new method for the rapid classification of rock mass around a tunnel during tunnel construction. [ABSTRACT FROM AUTHOR]

The design of a high-performance building necessitates tradeoffs among multiple performance objectives, and many simulation-based optimization tools have been developed for this purpose. In practice, these tools are often constrained by excessive computational load and tight project deadlines. This study aimed to develop a holistic approach to rapidly identify optimal building design schemes. A novel two-stage model was developed as a surrogate to conventional physics-based models, which were then linked to multi-objective optimization algorithms, namely, NSGA-II, NSGA-III, and C-TAEA, in search of the Pareto optimal and best design schemes. The performance of the two-stage model was further enhanced by applying the least absolute shrinkage and selection operator (LASSO) and Neural Architecture Search methods and learning from the statistical layer. The above approach was tested in the design of an apartment building for seniors in Northern China such as thermal comfort, carbon, and cost. The optimal design achieved through the compromise optimum can significantly reduce thermal discomfort, life-cycle carbon emissions, and high levels of daylighting conditions. The design scheme was visualized using a geometrical remodel module. This study contributes methodologically to complex multi-objective building optimization problems with practical implications for design. • This study uses LASSO for feature selection to improve ANN model performance. • A two-stage model is developed to improve prediction accuracy further. • Three optimization algorithms are deployed to ensure the quality of Pareto schemes. • Comfort, carbon emission, and cost are considered during optimization. • A remodeling module is developed to reform 3D building models from numerical schemes. [ABSTRACT FROM AUTHOR]

• Storage capacity estimation model is developed for national reservoirs in China. • Different statistical and machine learning models are compared and evaluated. • Storage capacity of all unrecorded Chinese reservoirs is estimated. • The spatial characteristics of reservoirs number and storage capacity are analyzed. Reservoir is essential for water resources management and utilization, and its storage capacity is the key information for the scientific and engineering community. Over 95 % of the ∼ 100,000 reservoirs constructed in China lack accurate capacity information due to the high cost and difficult accessibility of field measurements. Here we develop models of reservoir capacity estimation based on statistical and machine learning (ML) algorithms for the national-scale reservoirs in China. Based on the collected capacity records of a small portion of reservoirs and surrounding environment factors, we built statistical models by considering their associated variables, as well as ML models with different algorithms to predict the unknown reservoir capacities. The prediction results from different statistical and ML models were evaluated by comparing with the validated samples, and the optimal model was selected to estimate the reservoir capacity for further analyses. Results show that among the four statistical models and six ML models, the ML model using the extreme-gradient-boosting-tree (XGBoost) algorithm performs best in capacity prediction with a correlation coefficient of 0.97, a Nash-Sutcliffe efficiency of 0.95, and a mean absolute percentage error of 22.93 %. The estimated capacity of unrecorded reservoirs using XGBoost was 279.28 km3 (254.52–306.47 km3), and the total capacity of Chinese reservoirs was 1,065.19 km3 (1,040.43–1,092.38 km3). The proposed method provides a feasible approach of rapid estimation of reservoir capacity and monitoring of storage variations at large scales, which is critical for the rational conservation and utilization of water resources and the formulation of related policies. [ABSTRACT FROM AUTHOR]

• Develop a deep learning model (GWNN) for landslide susceptibility mapping. • Compare the performance of SVM, ANN, RF, GWR and GWNN models. • The proposed model can capture the spatial heterogeneity of landslides. • Reliable landslide susceptibility map in Yichang City using the proposed model. Landslides are among the most devastating natural disasters worldwide. Landslide susceptibility mapping (LSM) is a scientific approach for assessing landslides-prone areas. Despite the use of various statistical and machine learning (ML) models to predict landslide susceptibility, few have considered the effect of spatial location on landslides. Geographically weighted regression (GWR) is a common method for modeling spatial heterogeneity, but it may not be effective in solving complex nonlinear problems. This leads to inadequate prediction accuracy for landslide susceptibility at the local scale. This study introduces a geographically weighted neural network (GWNN) model to address this problem. The GWNN model combines the advantages of GWR and neural networks (NN) to assess the effect of landslide conditioning factors on landslides. The effectiveness of the proposed method was evaluated using landslide data collected from Yichang City, Hubei Province. The results show that GWNN exhibits the higher performance (AUC = 0.788) compared with traditional ML models and GWR. Furthermore, the superior ability of GWNN is demonstrated to capture spatial heterogeneity and predict landslide susceptibility accurately at local scales. The proposed method provides a valuable reference for modeling landslide-prone areas that exhibit spatial heterogeneity. [ABSTRACT FROM AUTHOR]

This paper has observed pivotal changes in innovative frameworks at the intersection of the digital revolution and sustainable initiatives. Within Industry 4.0, the fusion between intelligent manufacturing and green innovation has become a central driver for sustainability. Integrating these two pivotal forces to harness their collective efficiency is crucial. However, current literature offers limited insights into their combined impact, and the direction of their evolving trends still needs to be defined. Considering this, the research elucidates the core concepts and traits of digital-green fusion (DGF) within regional advanced manufacturing (RAMIDGF). Moreover, this research proposes a double-helix fusion mechanism model. Utilizing data from the advanced manufacturing sectors in six key Chinese urban clusters, this paper adopts a comprehensive analytical toolkit that includes the Coupling Modified Model, Kernel Density Estimation, Markov Chain algorithms, and the Dagum Gini Coefficient for trend analysis. The findings indicate that between 2009 and 2021, RAMIDGF exhibited a trend of initial growth followed by a decline, showing a strong correlation with per capita GDP (RGDP). The principal source of variation in RAMIDGF stems from differences between urban agglomerations, with the Yangtze River Delta and the Beijing-Tianjin-Hebei clusters exhibiting transitional behavior and a 'ladder-type trend' in RAMIDGF. As metropolitan clusters become more economically integrated, the probability of transitional behavior in RAMIDGF heightens. However, achieving elevated leapfrog development proves challenging, with an inherent risk of decline. Spatial spillover implications, agglomeration characteristics, and club convergence tendencies mark the RAMIDGF process. These findings expand upon new paradigms for regional sustainable development, offering novel avenues for realizing high-quality growth. [Display omitted] • The conceptual connotation and characteristics of RAMIDGF are proposed. • The double helix fusion model is constructed to analysis the fusion mechanism. • The RAMIDGF level of urban agglomerations is evaluated. • The trend and characteristics of RAMIDGF in urban agglomerations are demonstrated. [ABSTRACT FROM AUTHOR]

Quality control (QC) of marine data is a critical aspect in ensuring the usability of oceanic data. In this paper, we propose a novel QC method for time-series ocean wave data, which combines deep learning prediction and statistical analysis. Our method first realizes multi-element LSTM prediction of the time-series ocean wave observation data, capturing both temporal consistency and physical relationships between the multi-element inputs. Then, it applies peak detection on the regional mean difference ratio derived from the predicted and true values of the ocean wave data, and finally labels the anomalous data points based on peak detection results. We conducted experiments on the time-series wave data of four sites from the National Marine Science Center, China, and compared our proposed method with traditional QC method that is currently used for operational marine data quality control, as well as classic anomaly detection models including Isolation Forest and VAE-LSTM. The results show a significant improvement in the Precision, Recall, and F1 Score for erroneous samples using our proposed method. Our proposed method takes advantages of both deep learning and statistical analysis and considers physical correlation of multiple elements of marine data, effectively addressing the problem of erroneous discrimination of abnormal sea conditions in the traditional method, and providing valuable insights for the study of marine time-series observational data. • The results show a significant improvement in the Precision, Recall, and F1 Score for negative samples using our proposed approach. • By integrating deep learning and statistical methods, our proposed method achieves unsupervised quality control, effectively addressing the problem of erroneous discrimination of abnormal sea conditions in traditional methods, and providing valuable insights for the study of marine time-series observational data. [ABSTRACT FROM AUTHOR]

Do young children extract patterns from print to which they are naturally exposed but only in a limited way? This study investigated Chinese kindergarteners’ sensitivity to regularities in pinyin, an alphabetic coding system for non-alphabetic Chinese characters to which kindergarteners have limited exposure and little motivation for learning. Sixty 4-and 5-year-olds from Beijing took a pinyin decision task and a pinyin learning task and were assessed on Chinese word reading, Chinese word writing, and non-verbal IQ. Sensitivity to letter patterns in pinyin spellings and sensitivity to letter-sound correspondences in pinyin syllables appeared in 5-year-olds. After statistically controlling for age and IQ, sensitivity to letter-sound correspondences in pinyin syllables explained 4% additional variance in Chinese word reading and 7% additional variance in Chinese word writing. This study contributes novel evidence to statistical learning in early literacy, demonstrating that children are adept at pattern recognition in print even under limited conditions. [ABSTRACT FROM PUBLISHER]

Previous road transportation sustainability studies not only neglected the epistemic uncertainty that surrounds the impact of innovation and Research and Development (R&D) expenditure on pollutant emissions performance but also failed in designing and simultaneously exploring the strengths of alternative MCDM (Multiple-Criteria Decision-Making) approaches to better discriminate performance scores. This paper focuses on these two gaps by presenting a road transportation sustainability performance assessment of 29 Chinese provinces for a 14-year period in light of relevant socioeconomic and demographic variables. First, a novel TEA-IS model is developed to assess road transportation sustainability performance. Besides possessing the beneficial features of each model, this hybrid DEA-TOPSIS can analyze the sustainability performance from the perspective of the synergistic effects among the criteria. From the socioeconomic and demographic perspectives, we use machine learning techniques for predicting high-low performance and synergistic Chinese provinces. Results suggest that the discriminatory power of TEA-IS is good and there is high synergy in Chinese provinces in terms of sustainable road transportation. We further find that there is a high level of heterogeneity in road transportation sustainability among different geographical locations in China. The results further suggest that higher levels of synergy are strongly associated with medium and high-performing provinces. Finally, we find that innovation index, foreign direct investment, and non-coastal cities have both higher performance and synergy levels. We recommend that Chinese government should further provide favorable policies to enhance innovation and efforts should also be made to provide better environment for attracting more foreign direct investment. Finally, instead of road transportation, other modes of transportation, such as air and sea, are recommended to be further developed and utilized. • a TEA-IS model is developed to assess road transportation sustainability performance. • Machine learning methods are employed to predict high-low performance. • There is a high level of heterogeneity in road transportation sustainability in China. • Innovation and foreign direct investment are positively related to performance. [ABSTRACT FROM AUTHOR]

It will be beneficial to devise an effective approach for short-term macroeconomic forecasting. Existing traditional statistics-based macroeconomic forecasting mainly focuses on exploring feasible methods for improving the accuracy of long-term predictions. However, the performance of short-term predictions was far less impressive under the impact of unexpected incidents. Furthermore, some deep learning-based approaches can achieve fine-grained variable frequency forecasting and preliminarily demonstrate effective results, but the interpretability is still controversial. Therefore, how to consider both the performance and the interpretability has already become a universal concern and an urgent unsolved problem. In this paper, we identified the above issue and proposed an interpretable data-driven approach, named EcoForecast 1 1 The code and data are available at https://github.com/navfour/ecoforecast. , for short-term macroeconomic forecasting based on the N-BEATS (neural basis expansion analysis for interpretable time series forecasting) neural network. To the best of our knowledge, EcoForecast is the first interpretable purely data-driven unified normative scheme for macroeconomic forecasting that achieves variable forecast frequencies and prediction domains, surpassing traditional statistics-based and deep learning-based approaches in performance or interpretability. EcoForecast used a three-level hierarchical signal encoding, including the fully connected neural network (FCNN) level, the block level, and the stack level. The FCNN level implemented both forecast and backcast information extraction for the temporal prediction and parameter learning of context. Block levels were connected by residuals so that the block's backcast could be sequentially filtered on the input. The stack level was used to form the top-level system of EcoForecast, where each stack was constrained to specialize in different inductive functions. To some extent, EcoForecast can balance effectiveness, efficiency, generalizability and interpretability while conditions such as forecast frequency and time window change, even when unexpected incidents occur. Based on the actual macroeconomic data for China from 1992 to 2022, the data-driven EcoForecast demonstrated high stability in different sequence learning scenarios and the accompanying high-accuracy performance. This stability was reflected in smaller prediction error expectation and variance, tolerance of fewer input samples, and robustness across prediction domains. The experimental results indicated that EcoForecast improved the accuracy up to 3.94 times compared with the traditional BVAR. In the robustness test, EcoForecast required only a quarter of the data to achieve 2.51 times smaller forecast errors than the BVAR while also improving the accuracy of varied macroeconomic indicators such as the Purchasing Managers' Index (PMI) and national electricity generation (ELEC)forecasting by 2.38 and 1.45 times. EcoForecast had a high sensitivity to the emergence of economic inflection points and adapts quickly when the economic environment changes, thus demonstrating a performance that exceeds traditional solutions in GDP forecasting during epidemics and PMI forecasting during economic turmoil. Interfacing with traditional economics research, the interpretable EcoForecast can uncover the trends and cycles of economic change, from which the conclusions are validated with the actual economics practice in China. Our findings can provide a new possible research direction for short-term macroeconomic forecasting. [ABSTRACT FROM AUTHOR]

Regional eco-efficiency affects local public health through intermediaries such as economic and environmental impacts. Considering multiple factors, the implicit and uncertain relationship with regional characteristics, and the limited data availability, this paper investigated the forecasting of changes in local public health-including the number of visits to hospitals (VTH), outpatients with emergency treatment (OWET), number of inpatients (NI), number of health examinations (NOHE), and patients discharged (PD)-using calculated regional eco-efficiency with the Least Square-Support Vector Machine-Forecasting Model and acquired empirical evidence, utilizing the province-level data in China. Results: (1) regional eco-efficiency is a good predictor in both a single and multi-factor situation; (2) the prediction accuracy for five dimensions of the changes in local public health was relatively high, and the volatility was lower and more stable throughout the whole forecasting period; and (3) regional heterogeneity, denoted by three economic and demographic factors and three medical supply and technical level factors, improved the forecasting performance. The findings are meaningful for provincial-level decision-makers in China in order for them to know the current status or trends of medical needs, optimize the allocation of medical resources in advance, and enable ample time to tackle urgent emergencies, and, finally, the findings can serve to evaluate the social effects of improving regional eco-efficiency via local enterprises or individuals and adopting sustainable development strategies.

• A decomposition-ensemble framework is built for single-parameter modeling of superposed periodic signals. • An augmented lncosh ridge regression is proposed to handle the outliers and consider the static settling of suspended sediments. • A case study shows the framework realized 6-hour ahead SSC forecast with MREs of 5.80–9.44%. [Display omitted] An in-situ monitoring of water quality (suspended sediment concentration, SSC) and concurrent hydrodynamics was conducted in the subaqueous Yellow River Delta in China. Empirical mode decomposition and spectral analysis on the SSC time series reveal the different periodicities of each physical mechanism that contribute to the SSC variations. Based on this physical understanding, the decomposed SSC time series were trained separately with a newly-proposed augmented lncosh ridge regression, in which (1) a lncosh function was incorporated in traditional ridge regression for handling outliers in original data, and (2) the temporal auto-correlation in the decomposed SSC series was used for augmented regression. Finally, the trained sub-series were added up as the final prediction. The advantages of this decomposition-ensemble framework is that it depends on SSC only, superior to the normal process-based models which need the concurrent hydrodynamics for estimating bed shear stress. This will not only reduce the measurement uncertainties of the input when training the data-driven model, but also save the prediction cost as no other parameters than SSC need to be measured and input for running the model. The framework realized 6-hour-ahead high-accuracy forecasting with mean relative errors of 5.80–9.44% in the present case study. The proposed framework can be extended to forecast any signal that is superposed by components with various timescales (periodicities) which is common in nature. [ABSTRACT FROM AUTHOR]

To gain insight on the meteorological effects of air pollution, we study the relative importance of surface meteorological variables and boundary layer height (BLH) on six major air pollutants according to the orders of variables being selected in a forward variable selection algorithm. It is found that there was a strong agreement in the orders of relative importance for the major pollutants among six major cities in North China, which implies regularities in the meteorological processes of air pollution in that region. In particular, PM 2.5 , PM 10 , SO 2 and CO shared a common variable importance order and were mostly impacted by the dew point temperature and air pressure, while the NO 2 and O 3 were mostly influenced by the boundary layer height (BLH) and temperature, respectively. We evaluate the impacts of BLH on the pollution levels given the surface meteorological variables. It is found that BLH can be well modeled by the surface meteorological variables. Thus, air quality assessment without using BLH would also produce adequate results. • Stable orders of meteorological effects on six air pollutants in North China. • Boundary layer height's effects can be modeled by the surface meteorological variables. • PM 2.5 , PM 10 , SO 2 and CO shared a common order of variable importance. [ABSTRACT FROM AUTHOR]

Because the equation for calculating the air quality index (AQI) only considers the most serious pollutants, it remains controversial, and a comprehensive AQI has been a focus of subsequent studies. The present research transformed 22,504,440 pollution characteristics per year into datasets of a Louvain community detection clustering analysis. Unsupervised machine learning was applied to classify 367 cities across China into seven categories. The seven representative cities were selected from seven categories, and the augmented Dickey-Fuller test was used to detect their stationarity. A situation-based composite AQI calculation method was proposed. To verify the method, we clarified the difference between it and the AQI. We also compared its relative error with the total AQI. The relative error values in the seven representative cities were 8.9%, 12.16%, 12.91%, 8.75%, 13.42%, 11.41%, and 11.27%. These values were smaller than the total AQI. The median values of I A Q I C O , I A Q I N O 2 , I A Q I S O 2 , I A Q I P M 2.5 , a n d I A Q I P M 10 remained relatively stable. Only the changes in I A Q I O 3 were dramatic. High T h o u r values are generally encountered in more polluted cities. When there is at most one chief pollutant (CP), then T h o u r = 1. When there are several CPs, T h o u r > 1. The findings of this research provide the public with an intuitive understanding of air pollution and guidance on effective assessments of outdoor air quality. [Display omitted] • A situation-based CAQI method was proposed based on the number of I A Q I P values greater than 50. • The relative error of CAQI relied on air quality and the coefficients of chief pollutants (CPs) and sub-CPs. • A high T h o u r value was usually encountered in polluted cities with several CPs. • The 24-h median values of the individual air quality index remained relatively stable. • The situation-based CAQI may be used to identify the amount of pollutants. [ABSTRACT FROM AUTHOR]

The increasing release of nutrients to aquatic environments has led to great concern regarding eutrophication and the risk of unwanted algal blooms. Based on observational data of 20 water quality parameters measured on a monthly basis at 40 stations from 2011 to 2020, this study applied different Machine Learning (ML) algorithms to suggest the best option for algal bloom prediction in the Han River, a large river in South Korea. Eight different ML algorithms were categorized into several groups of statistical learning, regression family, and deep learning, and were then compared for their suitability to predict the chlorophyll-derived trophic index (TSI-Chla). ML algorithms helped identify the most important water quality parameters contributing to algal bloom prediction. The ML results confirmed that eutrophication and algal proliferation were governed by the complex interplay between nutrients (nitrogen and phosphorus), organic contaminants, and environmental factors. Of the models tested, the adaptive neuro-fuzzy inference system (ANFIS) exhibited the best performance owing to its consistent and outperforming prediction both quantitatively (i.e., via regression) and qualitatively (i.e., via classification), which was evidenced by the lowest value of mean absolute error (MAE) of 0.09, and the highest F1-score, Recall and Precision of 0.97, 0.98 and 0.96, respectively. In a further step, a representative web application was constructed to assist common users to predict the trophic status of the Han River. This study demonstrated that ML techniques are not only promising for highly accurate water quality modeling of urban rivers, but also reduce time and labor intensity for experiments, which decreases the number of monitored water quality parameters, providing further insights into the driving factors of water quality deterioration. They ultimately help devise proactive strategies for sustainable water management., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2021 Elsevier B.V. All rights reserved.)

To identify the water-use characteristics of sectors, previous studies primarily focused on measuring water use from the production side and the consumption side. In addition to sectors causing pressure on water resources as producers and drivers, there are also other important water-using sectors as primary suppliers and intermediate good suppliers. Income-based and betweenness-based methods are introduced here to measure the water use from these viewpoints. Using China as a case study, the results reveal critical differences between different perspectives, and sectors with novel water use characteristics have been identified. The K-means clustering method is used to cluster 139 sectors into five categories with distinctive features, of which the intermediate production drivers (category 2) are the critical virtual water transmission sectors which are rarely focused on before. New water-saving strategies can be revealed from the income perspective and the betweenness perspective, such as levying an income tax or improving the use efficiency of upstream input. The novelty of this study lies in the application of an unsupervised statistical learning technique to synthesize the different roles played by economic sectors for water-saving. • The water use of 139 sectors in 2012 China is measured from four perspectives. • The income-based and betweenness-based methods reveal critical but often neglected sectors. • The 139 sectors are clustered into five categories with distinctive characteristics. • Targeted sector-specific water-saving suggestions are offered. [ABSTRACT FROM AUTHOR]

• Three types of data-based landslide susceptibility prediction (LSP) models are compared. • Machine learning has higher LSP accuracy than heuristic and general statistical models. • Accurate landslide susceptibility maps of Shicheng County are produced. • Contributions of different factors to susceptibility prediction are identified. • Distribution rules of landslide susceptibility indexes of each model are explored. Commonly used data-driven models for landslide susceptibility prediction (LSP) can be mainly classified as heuristic, general statistical or machine learning models. This study plans to compare the prediction performance of these data-driven models on the landslide susceptibility mapping, thus further to explore the inherently features of these data-driven models. As a result, a more accurate and reliable LSP can be realized through choosing an optimal data-based model. A heuristic model represented by the analytic hierarchy process (AHP), a general statistical model represented by the general linear model (GLM) and information value (IV) model, and machine learning models represented by binary logistic regression (BLR), Multilayer Perceptron (MLP), back-propagation neural network (BPNN), support vector machine (SVM) and C5.0 decision tree (C5.0 DT) are adopted in this study. Shicheng County in China is used as the study area. In total, 369 landslides identified through field investigation are classified as training (70%) and testing datasets (30%). Next, 13 landslide conditioning factors (elevation, slope, aspect, plan curvature, profile curvature, relief amplitude, total surface radiation, population density, Normalized difference vegetation index, distance to river, topographic wetness index and rock types) are acquired from data sources of the free remote sensing images, Digital Elevation Model, field investigation and government reports. The correlations between these conditioning factors and the landslide locations are determined by frequency ratio analysis. Then, the landslide susceptibility indexes (LSIs) calculated by the eight trained models are imported into GIS software to produce landslide susceptibility maps of Shicheng County. Finally, the area under receiver operating characteristic curve (AUC), the calculated LSIs are applied to assess the LSP performance of the present eight models. The testing results show that these eight models generate reasonable LSP results as a whole, further showing that the C5.0 DT is of the highest prediction accuracy with an AUC value of 0.868, followed by the SVM (0.813), BPNN (0.803), MLP (0.792), BLR (0.784), GLM (0.779), IV (0.774) and AHP (0.773). It can be inferred that the machine learning models have higher LSP performance than general statistical and heuristic models due to its high AUC accuracy and reasonable LSIs distribution features, while general statistical model is limited by its linear analysis and heuristic model is limited by subjective weighting process. [ABSTRACT FROM AUTHOR]

• An end-to-end model is proposed for rice yield prediction using deep learning fusion. • Deep spatial-temporal features significantly reduced prediction errors. • The model converges quickly to low errors and provides stable prediction performance. • The model achieves the optimal prediction performance when the number of layers is 6. Rice yield is essential for more than half of the world's population, and thus, accurate predictions of rice yield are of great importance for trade, development policies, humanitarian assistance, decision-makers, etc. However, traditional mechanistic models and statistical machine learning models need to identify features, making the research on and application of these models laborious and time-consuming. In this paper, a novel end-to-end prediction model that fuses two back-propagation neural networks (BPNNs) with an independently recurrent neural network (IndRNN), named BBI-model, is proposed to address these challenges. In stage one, BBI-model preprocesses the original area and meteorology data. In stage two, one BPNN and the IndRNN are used to learn deep spatial and temporal features in parallel. In stage three, another BPNN combines two kinds of deep features and learns the relationships between these deep features and rice yields to make predictions for summer and winter rice yields. The experimental results indicate that BBI-model achieved the lowest mean absolute error (MAE) and root mean square error (RMSE) for the summer rice prediction (0.0044 and 0.0057, respectively) and corresponding values of 0.0074 and 0.0192 for the winter rice prediction when the number of layers in the network was set to six. Moreover, the errors of the model using the combination of deep spatial-temporal features were significantly lower than when simply using deep temporal features. Furthermore, the model converged quickly with 100 iterations and then remained stable. These findings confirm that the model can make accurate predictions for summer and winter rice yields of 81 counties in the Guangxi Zhuang Autonomous Region, China. [ABSTRACT FROM AUTHOR]

This paper is concerned with the statistical learning of the extreme smog (PM 2.5 ) dynamics of a vast region in China. Differently from classical extreme value modeling approaches, this paper develops a dynamic model of conditional, exponentiated Weibull distribution modeling and analysis of regional smog extremes, particularly for the worst scenarios observed in each day. To gain higher modeling efficiency, weather factors will be introduced in an enhanced model. The proposed model and the enhanced model are illustrated with temporal/spatial maxima of hourly PM 2.5 observations each day from smog monitoring stations located in the Beijing–Tianjin–Hebei geographical region between 2014 and 2019. The proposed model performs more precisely on fittings compared with other previous models dealing with maxima with autoregressive parameter dynamics, and provides relatively accurate prediction as well. The findings enhance the understanding of how severe extreme smog scenarios can be and provide useful information for the central/local government to conduct coordinated PM 2.5 control and treatment. For completeness, probabilistic properties of the proposed model were investigated. Statistical estimation based on the conditional maximum likelihood principle is established. To demonstrate the estimation and inference efficiency of studies, extensive simulations were also implemented. [ABSTRACT FROM AUTHOR]

• A strategy of the Bayesian framework was developed to guide Microcystis blooms and microcystin risk. • Total phosphorus influences the biomass of Microcystis , whereas total nitrogen determines the risk of microcystin. • The nutrient criteria are crucial for reducing the risk of microcystin. • Critical nutrient thresholds are lake-dependant and counteracted by temperature warming. The frequency of toxin-producing cyanobacterial blooms has increased in recent decades due to nutrient enrichment and climate change. Because Microcystis blooms are related to different environmental conditions, identifying potential nutrient control targets can facilitate water quality managers to reduce the likelihood of microcystins (MCs) risk. However, complex biotic interactions and field data limitations have constrained our understanding of the nutrient-microcystin relationship. This study develops a Bayesian modelling framework with intracellular and extracellular MCs that characterize the relationships between different environmental and biological factors. This model was fit to the across-lake dataset including three bloom-plagued lakes in China and estimated the putative thresholds of total nitrogen (TN) and total phosphorus (TP). The lake-specific nutrient thresholds were estimated using Bayesian updating process. Our results suggested dual N and P reduction in controlling cyanotoxin risks. The total Microcystis biomass can be substantially suppressed by achieving the putative thresholds of TP (0.10 mg/L) in Lakes Taihu and Chaohu, but a stricter TP target (0.05 mg/L) in Dianchi Lake. To maintain MCs concentrations below 1.0 μg/L, the estimated TN threshold in three lakes was 1.8 mg/L, but the effect can be counteracted by the increase of temperature. Overall, the present approach provides an efficient way to integrate empirical knowledge into the data-driven model and is helpful for the management of water resources. [ABSTRACT FROM AUTHOR]

Background: In countries where air pollution stations are unavailable or scarce, station measurements from other countries and atmospheric remote sensing could jointly provide information to estimate ambient air quality at a sufficiently fine resolution to study the relationship between air pollution exposure and health. Predicting NO 2 concentration globally with sufficient spatial and temporal resolution and accuracy for health studies is, however, not a trivial task. Challenges are data deficiency, in terms of NO 2 measurements and NO 2 predictors, and the development of a statistical model that can typify the regional and continental differences, such as traffic regulations, energy sources, and local weather., Objective: We investigated the feasibility of mapping daytime and nighttime NO 2 globally at a high spatial resolution (25 m), by including TROPOMI (TROPOspheric Monitoring Instrument) data and comparing various statistical learning techniques., Method: We separated daytime (7:00 am - 9:59 pm) and nighttime (10:00 pm - 6:59 am) based on the local times. To study if one should build models for each country separately, national models in 4 selected countries (the US, China, Germany, Spain) were developed. We build the models for 2017 and used 3636 stations. Seven statistical learning techniques were applied and the impact of the predictors, model fitting, and predicting accuracy was compared between different techniques, national models, national and global models, and models with and without including the NO 2 vertical column density retrieved from TROPOMI., Result and Conclusion: The ensemble tree-based methods obtained higher accuracy compared to the linear regression-based methods in national and global models. The global tree-based methods obtained similar accuracy to national models. Different spatial prediction patterns are observed even when the prediction accuracy is very similar. Separating between day and night can be important for more accurate air pollution exposure assessment. The TROPOMI variable is ranked as one of the most important variables in the statistical learning techniques but adding it to global models that contain other precedent remote sensing products does not improve the prediction accuracy., (Copyright © 2020 The Authors. Published by Elsevier Ltd.. All rights reserved.)

We examined the neural correlates of the statistical learning of orthographic-semantic connections in Chinese adult learners. Visual event-related potentials (ERPs) were recorded while participants were exposed to a sequence of artificial logographic characters containing semantic radicals carrying low, moderate, or high levels of semantic consistency. The behavioral results showed that the mean accuracy of participants' recognition of previously exposed characters was 63.1% that was significantly above chance level (50%), indicating the statistical learning of the regularities of semantic radicals. The ERP data revealed a temporal sequence of the neural process of statistical learning of orthographic-semantic connections, and different brain indexes were found to be associated with this processing, i.e., a clear N170-P200-N400 pattern. For N170, the larger negative amplitudes were evoked by the high and moderate consistency than the low consistency. For P200, the mean amplitudes elicited by the moderate and low consistency were larger than the high consistency. In contrast, a larger N400 amplitude was observed in the low than moderate and high consistency; and more negative amplitude was elicited by the moderate than high consistency. We propose that the initial potential shifts (N170 and P200) may reflect orthographic or graphic form identification, while the later component (N400) may be associated with semantic information analysis.

Background: The probability of heart failure during the perioperative period is 2% on average and it is as high as 17% when accompanied by cardiovascular diseases in China. It has been the most significant cause of postoperative death of patients. However, the patient is managed by the flow of information during the operation, but a lot of clinical information can make it difficult for medical staff to identify the information relevant to patient care. There are major practical and technical barriers to understand perioperative complications.Methods: In this work, we present three machine learning methods to estimate risks of heart failure, which extract intraoperative vital signs monitoring data into different modal representations (statistical learning representation, text learning representation, image learning representation). Firstly, we extracted features of vital signs monitoring data of surgical patients by statistical analysis. Secondly, the vital signs data is converted into text information by Piecewise Approximate Aggregation (PAA) and Symbolic Aggregate Approximation (SAX), then Latent Dirichlet Allocation (LDA) model is used to extract text topics of patients for heart failure prediction. Thirdly, the vital sign monitoring time series data of the surgical patient is converted into a grid image by using the grid representation, and then the convolutional neural network is directly used to identify the grid image for heart failure prediction. We evaluated the proposed methods in the monitoring data of real patients during the perioperative period.Results: In this paper, the results of our experiment demonstrate the Gradient Boosting Decision Tree (GBDT) classifier achieves the best results in the prediction of heart failure by statistical feature representation. The sensitivity, specificity and the area under the curve (AUC) of the best method can reach 83, 85 and 84% respectively.Conclusions: The experimental results demonstrate that representation learning model of vital signs monitoring data of intraoperative patients can effectively capture the physiological characteristics of postoperative heart failure. [ABSTRACT FROM AUTHOR]

• Chlorophyll a predicted by statistical machine learning successfully displayed seasonal variations. • Estimated nutrient thresholds were quite variable and higher in summer than in winter. • Reducing TP concentration in these lakes in summer is important to control algal growth. • Seasonal variation in nutrient concentration should be considered when establishing criteria. Eutrophication and subsequent harmful cyanobacteria blooms are global water quality problems, and identifying the key drivers of water eutrophication and estimating nutrient thresholds for it in waterbodies have long been challenges for water quality managers. Data-intensive machine learning models have been shown to be better able to reveal the nonlinear relationships between variables in the study of complex biotic community dynamics than traditional mechanistic models. In this study, we applied random forest models to long-term datasets from nutrient monitoring and meteorological observations to characterize the relationships between algal growth and different environmental drivers in three eutrophic lakes in China. We further attempted to estimate the season-specific nutrient thresholds in these lakes, and assess the potential decreases in chlorophyll a concentrations that could be achieved through nutrient management. In general, chlorophyll a concentrations predicted by the random forest models were consistent with the values observed in the lakes, and successfully displayed the same seasonal variations. The estimated total nitrogen (TN) and total phosphorus (TP) nutrient thresholds were quite variable among months, and were higher in summer than in winter. To maintain chlorophyll a concentrations below 20 μg/L, the estimated TN thresholds in Lakes Taihu, Dianchi, and Chaohu in August were 2145 ± 683, 2372 ± 918 and 1527 ± 71 μg/L (mean ± standard deviation), respectively, and the corresponding TP thresholds were 82 ± 24, 149 ± 22, and 120 ± 22 μg/L. The modelling results indicated that it was more important to control the TP concentrations in these lakes than the TN concentrations to control algal growth in summer. In summary, the strong seasonal variation in the estimated nutrient thresholds suggests that a 'one-size-fits-all' nutrient control target could overprotect these water bodies. Seasonal variation in nutrient concentrations and environmental drivers should thus be considered when establishing nutrient criteria and setting nutrient control targets. [ABSTRACT FROM AUTHOR]

Flash floods in mountainous catchments are often caused by the rainstorm, which may result in more severe consequences than plain area floods due to less timescale and a fast-flowing front of water and debris. Flash flood forecasting is a huge challenge for hydrologists and managers due to its instantaneity, nonlinearity, and dependency. Among different methods of flood forecasting, data-driven models have become increasingly popular in recent years due to their strong ability to simulate nonlinear hydrological processes. This study proposed a Support Vector Regression (SVR) model, which is a powerful artificial intelligence-based model originated from statistical learning theory, to forecast flash floods at different lead times in a small mountainous catchment. The lagged average rainfall and runoff are identified as model input variables, and the time lags associated with the model input variables are determined by the hydrological concept of the time of response. There are 69 flash flood events collected from 1984 to 2012 in a mountainous catchment in China and then used for the model training and testing. The contribution of the runoff variables to the predictions and the phase lag of model outputs are analyzed. The results show that: (i) the SVR model has satisfactory predictive performances for one to three-hours ahead forecasting; (ii) the lagged runoff variables have a more significant effect on the predictions than the rainfall variables; and (iii) the phase lag (time difference) of prediction results significantly exists in both two- and three-hours-ahead forecasting models, however, the input rainfall information can assist in mitigating the phase lag of peak flow. [ABSTRACT FROM AUTHOR]