Search

Based on authorized patents of China's artificial intelligence industry from 2013 to 2022, this paper constructs an Industry–University–Research institution (IUR) collaboration network and an Inter-Firm (IF) collaboration network and used the entropy weight method to take both the quantity and quality of patents into account to calculate the innovation performance of firms. Through the hierarchical clustering algorithm and classification and regression trees (CART) algorithm, in-depth analysis has been conducted on the intricate non-linear influence mechanisms between multiple variables and a firm's innovation performance. The findings indicate the following: (1) Based on the network centrality (NC), structural hole (SH), collaboration breadth (CB), and collaboration depth (CD) of both IUR and IF collaboration networks, two types of focal firms are identified. (2) For different types of focal firms, the combinations of network characteristics affecting their innovation performance are various. (3) In the IUR collaboration network, focal firms with a wide range of heterogeneous collaborative partners can obtain high innovation performance. However, focal firms in the IF collaboration network can achieve the same aim by maintaining deep collaboration with other focal firms. This paper not only helps firms make scientific decisions for development but also provides valuable suggestions for government policymakers. [ABSTRACT FROM AUTHOR]

Projecting precipitation changes is essential for researchers to understand climate change impacts on hydrological cycle. This study projected future precipitation over the Hanjiang River Basin (HRB) based on the multimodel ensemble (ME) of six global climate models from Phase 6 of the Coupled Model Intercomparison Project (CMIP6). An ME method using the fusion of four machine learning (ML) algorithms (random forest [RF], K‐nearest neighbors [KNN], extra tree [ET], and gradient boosting decision tree [GBDT]) was proposed in this study. The future precipitation changes were investigated during 2023–2042 (Near‐term), 2043–2062 (Mid‐term), and 2081–2100 (Long‐term) periods, with reference to the base period 1995–2014, under three integrated scenarios (SSP1‐2.6, SSP2‐4.5, and SSP5‐8.5) of the Shared Socioeconomic Pathways (SSPs) and the representative concentration pathways (RCPs). The results show that: (1) the proposed ME method performs better than the ME mean and individual ML algorithms, with a correlation coefficient value reaching 0.88 and Taylor skill score reaching 0.764. (2) The precipitation under SSP5‐8.5 has the largest upward trend with the annual precipitation variation range of −9.27% to 112.84% from 2023 to 2100, followed by SSP2‐4.5 with −30.48% to 44.67%, and the smallest under SSP1‐2.6 with −37.19% to 37.78%, which show a significant trend of humidification over the HRB in the future. (3) The precipitation changes over the HRB are projected to increase over time, with the largest in the Long‐term, followed by Mid‐term, and the smallest in the Near‐term. (4) The northeastern parts of the HRB are projected to experience a large precipitation in the future, and the southeastern parts are smaller. (5) Uncertainties in the projected precipitation over the HRB still exist, which can be reduced by ME. The findings obtained in this study have important implications for hydrological policymakers to make adaptive strategies to reduce the risks of climate change. [ABSTRACT FROM AUTHOR]

It is important to improve the accuracy of models estimating aboveground biomass (AGB) in large areas with complex geography and high forest heterogeneity. In this study, k-nearest neighbors (k-NN), gradient boosting machine (GBM), random forest (RF), quantile random forest (QRF), regularized random forest (RRF), and Bayesian regularization neural network (BRNN) machine learning algorithms were constructed to estimate the AGB of four forest types based on environmental factors and the variables selected by the Boruta algorithm in Yunnan Province and using integrated Landsat 8 OLI and Sentinel 2A images. The results showed that (1) DEM was the most important variable for estimating the AGB of coniferous forests, evergreen broadleaved forests, deciduous broadleaved forests, and mixed forests; while the vegetation index was the most important variable for estimating deciduous broadleaved forests, the climatic factors had a higher variable importance for estimating coniferous and mixed forests, and texture features and vegetation index had a higher variable importance for estimating evergreen broadleaved forests. (2) In terms of specific model performance for the four forest types, RRF was the best model both in estimating the AGB of coniferous forests and mixed forests; the R2 and RMSE for coniferous forests were 0.63 and 43.23 Mg ha−1, respectively, and the R2 and RMSE for mixed forests were 0.56 and 47.79 Mg ha−1, respectively. BRNN performed the best in estimating the AGB of evergreen broadleaved forests; the R2 was 0.53 and the RMSE was 68.16 Mg ha−1. QRF was the best in estimating the AGB of deciduous broadleaved forests, with R2 of 0.43 and RMSE of 45.09 Mg ha−1. (3) RRF was the best model for the four forest types according to the mean values, with R2 and RMSE of 0.503 and 52.335 Mg ha−1, respectively. In conclusion, different variables and suitable models should be considered when estimating the AGB of different forest types. This study could provide a reference for the estimation of forest AGB based on remote sensing in complex terrain areas with a high degree of forest heterogeneity. [ABSTRACT FROM AUTHOR]

Climate change and human activities have reduced the area and degraded the functions and services of wetlands in China. To protect and restore wetlands, it is urgent to predict the spatial distribution of potential wetlands. In this study, the distribution of potential wetlands in China was simulated by integrating the advantages of Google Earth Engine with geographic big data and machine learning algorithms. Based on a potential wetland database with 46,000 samples and an indicator system of 30 hydrologic, soil, vegetation, and topographic factors, a simulation model was constructed by machine learning algorithms. The accuracy of the random forest model for simulating the distribution of potential wetlands in China was good, with an area under the receiver operating characteristic curve value of 0.851. The area of potential wetlands was 332,702 km2, with 39.0% of potential wetlands in Northeast China. Geographic features were notable, and potential wetlands were mainly concentrated in areas with 400–600 mm precipitation, semi-hydric and hydric soils, meadow and marsh vegetation, altitude less than 700 m, and slope less than 3°. The results provide an important reference for wetland remote sensing mapping and a scientific basis for wetland management in China. [ABSTRACT FROM AUTHOR]

The red turpentine beetle (Dendroctonus valens LeConte) has caused severe ecological and economic losses since its invasion into China. It gradually spreads northeast, resulting in many Chinese pine (Pinus tabuliformis Carr.) deaths. Early detection of D. valens infestation (i.e., at the green attack stage) is the basis of control measures to prevent its outbreak and spread. This study examined the changes in spectral reflectance after initial attacking of D. valens. We also explored the possibility of detecting early D. valens infestation based on spectral vegetation indices and machine learning algorithms. The spectral reflectance of infested trees was significantly different from healthy trees (p < 0.05), and there was an obvious decrease in the near-infrared region (760–1386 nm; p < 0.01). Spectral vegetation indices were input into three machine learning classifiers; the classification accuracy was 72.5–80%, while the sensitivity was 65–85%. Several spectral vegetation indices (DID, CUR, TBSI, DDn2, D735, SR1, NSMI, RNIR•CRI550 and RVSI) were sensitive indicators for the early detection of D. valens damage. Our results demonstrate that remote sensing technology could be successfully applied to early detect D. valens infestation and clarify the sensitive spectral regions and vegetation indices, which has important implications for early detection based on unmanned airborne vehicle and satellite data. [ABSTRACT FROM AUTHOR]

• Adasyn and t-SNE are used to pre-process the collected rockburst data. • SVM, BP, RBF, RF and ELM optimized by global optimization algorithm are used to predict rockburst intensity. • An improved D-S evidence theory rockburst prediction fusion model based on multiple machine learning algorithms is raised. • The fusion model predicts much better than single algorithm model in engineering application. Accurate prediction of rockburst hazards is an effective means to improve the development and utilization of underground space. When using a single machine learning algorithm for rockburst prediction, the reliability of the classification results can hardly be guaranteed. In this paper, a fusion model for rockburst prediction based on multiple machine learning algorithms is proposed in combination with D-S evidence theory. Six characteristic parameters (σ θ , σ c , σ t , SR , BR and Wet) were statistically analyzed for 304 sets of rockburst cases, and the data had outliers and class imbalances. The box plot method was used to detect replacement of a small number of outliers, and the Adasyn oversampling method eliminated the effect of data class imbalance. The visualization of the data by using the t-SNE method showed improved classifiability. SVM, BP, RBF, RF and ELM rockburst prediction models were developed with the six characteristic parameters as model inputs. In addition, the global optimization algorithms were used to find the optimal hyperparameters of the five machine learning algorithms. The accuracy of the optimized models was improved by 2.63%, 5.27%, 2.63%, 6.58% and 7.5% compared to the unoptimized models. Based on the prediction results, the precision was proposed as the fusion index. Additionally, based on improved D-S evidence theory, a fusion model for rockburst prediction was developed with five optimized machine learning algorithm models as base classifiers. Finally, the fusion model was applied to the rockburst prediction of Jiangbian Hydropower Station and Sanshan Island Gold Mine in China with the accuracy of 92.86%, and the prediction was outperformed by a single base classifier. The fusion model compensated for the uncertainty and poor robustness of the prediction results of the single-base classifier, and improved the accuracy and reliability of rockburst prediction. [ABSTRACT FROM AUTHOR]

Precipitation prediction is central in hydrology and water resources planning and management. This paper introduces a semi-empirical predictive model to predict monthly precipitation and compares its predictive skill with those of machine learning (ML) methods. The stochastic method presented herein estimates monthly precipitation with one-step-ahead prediction properties. The ML predictive skill of the algorithms is evaluated by predicting monthly precipitation relying on the statistical association between precipitation and environmental and topographic factors. The semi-empirical predictive model features non-negative matrix factorization (NMF) for investigating the influence of multiple predictor variables on precipitation. The semi-empirical predictive model's parameters are optimized with the hybrid genetic algorithm (GA) and Levenberg-Marquardt algorithm (LM), or GALMA, yielding a validated model with high predictive skill. The methodologies are illustrated with data from Hubei Province, China, which comprise 27 meteorological station datasets from 1988–2017. The empirical results provide valuable insights for developing semi-empirical rainfall prediction models. [ABSTRACT FROM AUTHOR]

Reserving grains is important to safeguard food supply and a major prerequisite in national security. Grain quality changes during the reserving storage period, and an important grain quality indictor is fatty acid content. Understanding real-time change in fatty acid content of the grains is thus significant for safe storage of grains. Taking rice storage as an example, we present and compare four models for predicting change in its fatty acid content: multiple linear regression (MLR) model, artificial neural network (ANN) model, support vector regression (SVR) model, and least square support vector regression (LSSVR) model. Comparison of the four models was based on their coefficient of determination(R²), mean absolute error (MAE), mean absolute relative error and root mean square error (RMSE), in reproducing the observed change of fatty acid content. A total of 201 rice storage data were collected from 35 granaries at 5 grain depots from three provinces in the northeastern China. Each data set included inception of the rice storage, initial moisture, initial fatty acid content, moisture, effective accumulated temperature, duration of the storage, grain temperature, granary temperature, time at which the measurements were taken, and fatty acid content. The correlation between the fatty acid content and other parameters was analyzed using the Pearson correlation coefficient. Because of possible correlation between parameters, we reduced the number of the parameters using the principal components analysis by keeping only the key independent parameters, which were initial moisture, initial fatty acid content, effective accumulated temperature and grain temperature. These four parameters were normalized first, and we then randomly selected 80% of the data to train the models, with the remaining 20% to test the models. The particle swarm optimization (PSO) algorithm was used to optimize the parameters of the SVR and LSSVR models prior to the simulation. The model testing results showed that the coefficient of determinant, MAE, MAPE and RMSE of the LSSVR model was 0.911, 0.275 mg/100 g, 1.604% and 0.348 mg/100 g, respectively, significantly better than those of the ANN and SVR models and slightly better than that of the MLR model. A number of testing indicators revealed that the LSSVR and MLR models were most accurate while the SVR model was least accurate for predicting the fatty acid content in the rice. It can be concluded that the LSSVR and MLR models were accurate and reliable, and can be used to estimate change in fatty acid content of the rice using other easy-to-measure parameters. It has implications for estimating other quality indicators of grains in reserving storages. [ABSTRACT FROM AUTHOR]

Aiming to aid Electric-Power Industry to accurately understand users, hybrid learning model based ensemble learning algorithms for recognizing user to be sensitive to electric charge is proposed in this paper. On the basis of big data presented by CCF competition sponsor in China, with some excellent technology or algorithm such as JieBa, SFFS, etc., we extract many key features from data set and successfully draw a portrait for users who pay close attention to electric charge. Furthermore, machine learning algorithms and the strategy selection model related to them are investigated. The feasibility that hybrid learning model combining several ensemble learning algorithms can substantially improve classification accuracy are proved from theoretical level. Then the details of implementing hybrid learning model are described in the paper. Lastly, the hybrid learning model named Stacking is achieved, which yields better performance in contrast to the state-of-the-art competitors. The experimental results indicate that Stacking has both high precision and recall with 0.8 and 0.85 respectively. Furthermore the F1 score of Stacking evaluation is 0.823. [ABSTRACT FROM AUTHOR]

Background: Electronic medical records provide large-scale real-world clinical data for use in developing clinical decision systems. However, sophisticated methodology and analytical skills are required to handle the large-scale datasets necessary for the optimisation of prediction accuracy. Myopia is a common cause of vision loss. Current approaches to control myopia progression are effective but have significant side effects. Therefore, identifying those at greatest risk who should undergo targeted therapy is of great clinical importance. The objective of this study was to apply big data and machine learning technology to develop an algorithm that can predict the onset of high myopia, at specific future time points, among Chinese school-aged children. Methods and findings: Real-world clinical refraction data were derived from electronic medical record systems in 8 ophthalmic centres from January 1, 2005, to December 30, 2015. The variables of age, spherical equivalent (SE), and annual progression rate were used to develop an algorithm to predict SE and onset of high myopia (SE ≤ −6.0 dioptres) up to 10 years in the future. Random forest machine learning was used for algorithm training and validation. Electronic medical records from the Zhongshan Ophthalmic Centre (a major tertiary ophthalmic centre in China) were used as the training set. Ten-fold cross-validation and out-of-bag (OOB) methods were applied for internal validation. The remaining 7 independent datasets were used for external validation. Two population-based datasets, which had no participant overlap with the ophthalmic-centre-based datasets, were used for multi-resource validation testing. The main outcomes and measures were the area under the curve (AUC) values for predicting the onset of high myopia over 10 years and the presence of high myopia at 18 years of age. In total, 687,063 multiple visit records (≥3 records) of 129,242 individuals in the ophthalmic-centre-based electronic medical record databases and 17,113 follow-up records of 3,215 participants in population-based cohorts were included in the analysis. Our algorithm accurately predicted the presence of high myopia in internal validation (the AUC ranged from 0.903 to 0.986 for 3 years, 0.875 to 0.901 for 5 years, and 0.852 to 0.888 for 8 years), external validation (the AUC ranged from 0.874 to 0.976 for 3 years, 0.847 to 0.921 for 5 years, and 0.802 to 0.886 for 8 years), and multi-resource testing (the AUC ranged from 0.752 to 0.869 for 4 years). With respect to the prediction of high myopia development by 18 years of age, as a surrogate of high myopia in adulthood, the algorithm provided clinically acceptable accuracy over 3 years (the AUC ranged from 0.940 to 0.985), 5 years (the AUC ranged from 0.856 to 0.901), and even 8 years (the AUC ranged from 0.801 to 0.837). Meanwhile, our algorithm achieved clinically acceptable prediction of the actual refraction values at future time points, which is supported by the regressive performance and calibration curves. Although the algorithm achieved balanced and robust performance, concerns about the compromised quality of real-world clinical data and over-fitting issues should be cautiously considered. Conclusions: To our knowledge, this study, for the first time, used large-scale data collected from electronic health records to demonstrate the contribution of big data and machine learning approaches to improved prediction of myopia prognosis in Chinese school-aged children. This work provides evidence for transforming clinical practice, health policy-making, and precise individualised interventions regarding the practical control of school-aged myopia. [ABSTRACT FROM AUTHOR]

Background: In China, dengue remains an important public health issue with expanded areas and increased incidence recently. Accurate and timely forecasts of dengue incidence in China are still lacking. We aimed to use the state-of-the-art machine learning algorithms to develop an accurate predictive model of dengue. Methodology/Principal findings: Weekly dengue cases, Baidu search queries and climate factors (mean temperature, relative humidity and rainfall) during 2011–2014 in Guangdong were gathered. A dengue search index was constructed for developing the predictive models in combination with climate factors. The observed year and week were also included in the models to control for the long-term trend and seasonality. Several machine learning algorithms, including the support vector regression (SVR) algorithm, step-down linear regression model, gradient boosted regression tree algorithm (GBM), negative binomial regression model (NBM), least absolute shrinkage and selection operator (LASSO) linear regression model and generalized additive model (GAM), were used as candidate models to predict dengue incidence. Performance and goodness of fit of the models were assessed using the root-mean-square error (RMSE) and R-squared measures. The residuals of the models were examined using the autocorrelation and partial autocorrelation function analyses to check the validity of the models. The models were further validated using dengue surveillance data from five other provinces. The epidemics during the last 12 weeks and the peak of the 2014 large outbreak were accurately forecasted by the SVR model selected by a cross-validation technique. Moreover, the SVR model had the consistently smallest prediction error rates for tracking the dynamics of dengue and forecasting the outbreaks in other areas in China. Conclusion and significance: The proposed SVR model achieved a superior performance in comparison with other forecasting techniques assessed in this study. The findings can help the government and community respond early to dengue epidemics. [ABSTRACT FROM AUTHOR]

Today, China is facing a very serious issue of Air Pollution due to its dreadful impact on the human health as well as the environment. The urban cities in China are the most affected due to their rapid industrial and economic growth. Therefore, it is of extreme importance to come up with new, better and more reliable forecasting models to accurately predict the air quality. This paper selected Beijing, Tianjin and Shijiazhuang as three cities from the Jingjinji Region for the study to come up with a new model of collaborative forecasting using Support Vector Regression (SVR) for Urban Air Quality Index (AQI) prediction in China. The present study is aimed to improve the forecasting results by minimizing the prediction error of present machine learning algorithms by taking into account multiple city multi-dimensional air quality information and weather conditions as input. The results show that there is a decrease in MAPE in case of multiple city multi-dimensional regression when there is a strong interaction and correlation of the air quality characteristic attributes with AQI. Also, the geographical location is found to play a significant role in Beijing, Tianjin and Shijiazhuang AQI prediction. [ABSTRACT FROM AUTHOR]

[Display omitted] • Epikarst thickness in depression is markedly greater than that on hillslope. • Spherical model fits for the interpolation of spatial structure of epikarst thickness. • Soil thickness is identified as the primary control on epikarst thickness. • ANNs model is suitable for predicting epikarst thickness based on RFE for variable selection. Epikarst, as a distinct layer of karst critical zone, plays a crucial role in water storage-transfer functions and biogeochemical processes. The epikarst thickness, which shows high spatial variability, is an important parameter that controls the dynamic characteristics of the water cycle. Due to the time-cost constraints in conventional field surveys and coarse resolution of remote sensing interpretation, empirical statistical models based on epikarst forming factors are an alternative to quantify the epikarst thickness. In this study, we investigated the spatial variability of epikarst thickness in Huanjiang, Guangxi of southwest China, based on electrical resistivity tomography with 36 sampling lines (1,658 sampling points) in a dolomite catchment. The controlling factors of epikarst thickness were defined based on selecting terrain-vegetation-soil variables with GLM (generalized linear model), RFE (recursive feature elimination) and PLSR (partial least squares regression) methods. Then, the best method for predicting epikarst thickness was selected based on the model with the highest accuracy. Results showed that the mean epikarst thickness was 6.32 m; the epikarst was significantly thicker in the depression (7.2 m) than on the hill slope (5.8 m), whereas there were no significant differences among slope positions. Geostatistical analysis showed that the spherical model reflected the strong spatial autocorrelation of epikarst thickness. Based on the prediction performance of the models, the terrain-vegetation variables alone could not explain epikarst thickness satisfactorily (explanation rate, 32.9 %; testing accuracy, R2 <0.5). By contrast, the modeling accuracy significantly increased (64.7 %, 0.7) when soil properties were considered as predictors. Based on validation criteria (R2 and RMSE), RFE for selecting variables and ANNs for prediction (0.814, 2.137 m) were the optimal model set to predict the epikarst thickness. This study provides a reference method for predicting epikarst thickness at a hillslope scale and the information would be helpful for hydrological catchment modeling. [ABSTRACT FROM AUTHOR]

• COVID-19 images were analyzed with a new preprocessing model. • In this study, feature extraction was performed from COVID-19 images according to RGB values. • The proposed model was tested by statistical methods. • The results of both normal data and feature-reduced data were compared. The COVID-19 pandemic spread all over the world, starting in China in late 2019, and significantly affected life in all aspects. As seen in SARS, MERS, COVID-19 outbreaks, coronaviruses pose a great threat to world health. The COVID-19 epidemic, which caused pandemics all over the world, continues to seriously threaten people's lives. Due to the rapid spread of COVID-19, many countries' healthcare sectors were caught off guard. This situation put a burden on doctors and healthcare professionals that they could not handle. All of the studies on COVID-19 in the literature have been done to help experts to recognize COVID-19 more accurately, to use more accurate diagnosis and appropriate treatment methods. The alleviation of this workload will be possible by developing computer aided early and accurate diagnosis systems with machine learning. Diagnosis and evaluation of pneumonia on computed tomography images provide significant benefits in investigating possible complications and in case follow-up. Pneumonia and lesions occurring in the lungs should be carefully examined as it helps in the diagnostic process during the pandemic period. For this reason, the first diagnosis and medications are very important to prevent the disease from progressing. In this study, a dataset consisting of Pneumonia and Normal images was used by proposing a new image preprocessing process. These preprocessed images were reduced to 15x15 unit size and their features were extracted according to their RGB values. Experimental studies were carried out by performing both normal values and feature reduction among these features. RGB values of the images were used in train and test processes for MLAs. In experimental studies, 5 different Machine Learning Algorithms (MLAs) (Multi Class Support Vector Machine (MC-SVM), k Nearest Neighbor (k-NN), Decision Tree (DT), Multinominal Logistic Regression (MLR), Naive Bayes (NB)). The following accuracy rates were obtained in train operations for MLAs, respectively; 1, 1, 1, 0.746377, 0.963768. Accuracy results in test operations were obtained as follows; 0.87755, 0.857143, 0.857143, 0.877551, 0.938776. [ABSTRACT FROM AUTHOR]

• The accuracy of UAV–LiDAR point cloud height and intensity variables in AGB of invasive mangrove is compared for the first time. • The AGB of mangroves invaded by typical estuaries in subtropical China was estimated by UAV–LiDAR. • We developed a LiDAR mangrove aboveground biomass retrieval system. • CBR ML algorithm shows good effect in estimating AGB of invasive mangrove; • AGB of the invasive mangrove is related to elevation of the beach surface and main tide. Quantitative assessment of aboveground biomass (AGB) and spatial distribution pattern of exotic mangrove plants (Sonneratia apetala) is of great significance for blue carbon management and ecological restoration in typical subtropical estuaries in China. Although unmanned aerial vehicle (UAV) light detection and ranging (LiDAR) has certain advantages in the investigation of the vertical three-dimensional structure of mangroves, the existing mangrove investigation results are mainly based on plot investigation method. Few scholars use machine learning (ML) method to estimate AGB of invasive Sonneratia apetala by combining plot investigation and LiDAR data. Therefore, on the basis of the height and intensity variables of UAV-LiDAR data, this study used four different ML algorithms, namely, xgboost regressor (XGBR), catboost regressor (CBR), light gradient boosting regressor (LGBR) and AdaBoost regressor (ABR), to estimate AGB of invasive mangrove. Then, the quantitative relationship between invasive mangrove biomass and hydrological unit was analysed. We found that CBR model had the highest accuracy in estimation of mangrove AGB (R2 = 0.7644, RMSE = 11.1725 Mg/ha), followed by XGBR model (R2 = 0.6759, 13.1053 Mg/ha). However, LGBR model (R2 = 0.3506, RMSE = 18.5510 Mg/ha) had poor fitting effect. The AGB of invasive mangroves showed a spatial distribution pattern of high in northwest and low in southeast, and its value ranged from 7.31 Mg/ha to 114.04 Mg/ha, with an average of 25.57 Mg/ha. The AGB of invasive mangroves was independent of the area size of the hydrological response unit but depended on the elevation of the beach surface and the distance from the main tidal ditch. This study demonstrates the feasibility of UAV-LiDAR remote sensing and CBR model in estimating AGB of invasive mangrove species, which can provide scientific basis and technical support for the assessment of invasive mangrove ecosystem and the protection of local mangrove tree species. [ABSTRACT FROM AUTHOR]

• Built the storm surge disaster dataset with 50 indicators and 132 events. • Scientific data preprocessing methods ensure the reliability of the model. • Rank the importance of 50 indicators; hazard indicators are more important. • The indicator screening renders model simplification and workload reduction. • Satisfy machine learning models of storm surge disaster loss estimation are obtained. Storm surge is the most severe marine disaster in China, affecting the whole coastal area. Estimating storm surge disaster loss (SSDL) is significant to disaster prevention, sustainability and decision-making. Taking 11 provincial administrative regions in the coastal areas of China as the study area, this paper estimated SSDL grades based on four machine learning (ML) algorithms. A total of 132 pieces of official open-source data of storm surge disasters were collected and divided into a cross-validation set (CV set) and a test set. First, a comprehensive indicator system was constructed from three perspectives, covering the hazard (16) of disaster-causing factors, the vulnerability (22) and resilience (12) of disaster-bearing bodies, including 50 indicators. A few data preprocessing methods are implemented to improve the model performance such as normalization, SMOTE, etc. Then, Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Logistic model tree (LMT), and K-star were applied to construct the estimation model of SSDL grades. Principal component analysis (PCA) and recursive feature elimination (RFE) are adopted for an intelligent screening of the indicators. Finally, the models' performance is compared through Precision, Recall, F1 score and Kappa metrics. The results show that scientific and efficient data preparation is a strong guarantee for the reliability and stability of the models. RFE is verified more suitable for indicator selection in this paper compared with PCA. The importance ranking of RFE enhances the interpretability of the ML model, which shows that the hazard indicator is the most important, the vulnerability indicator is the second, and the resilience indicator is the least. The 27-indicator K-star model, with advantages of accurate estimation, strong generalization, and less workload, is the optimal SSDL estimation model. The number of input indicators of the optimal SSDL estimation model is 27, its CV Precision, Recall, F1 score, and Kappa are 0.838, 0.832, 0.827, and 0.776, and its Precision, Recall, F1 score, and Kappa for test set are 0.819, 0.786, 0.781, and 0.714, respectively. This paper provides a scientific basis for the government's decision-making and risk management, and it can be used as a typical demonstration case of SSDL research. [ABSTRACT FROM AUTHOR]

Due to the country's rapid economic growth, the problem of air pollution in China is becoming increasingly serious. In order to achieve a win-win situation for the environment and urban development, the government has issued many policies to strengthen environmental protection. PM2.5 is the primary particulate matter in air pollution, so an accurate estimation of PM2.5 distribution is of great significance. Although previous studies have attempted to retrieve PM2.5 using geostatistical or aerosol remote sensing retrieval methods, the current rough resolution and accuracy remain as limitations of such methods. This paper proposes a fine-grained spatiotemporal PM2.5 retrieval method that comprehensively considers various datasets, such as Landsat 8 satellite images, ground monitoring station data, and socio-economic data, to explore the applicability of different machine learning algorithms in PM2.5 retrieval. Six typical algorithms were used to train the multi-dimensional elements in a series of experiments. The characteristics of retrieval accuracy in different scenarios were clarified mainly according to the validation index, R2. The random forest algorithm was shown to have the best numerical and PM2.5-based air-quality-category accuracy, with a cross-validated R2 of 0.86 and a category retrieval accuracy of 0.83, while both maintained excellent retrieval accuracy and achieved a high spatiotemporal resolution. Based on this retrieval model, we evaluated the PM2.5 distribution characteristics and hourly variation in the sample area, as well as the functions of different input variables in the model. The PM2.5 retrieval method proposed in this paper provides a new model for fine-grained PM2.5 concentration estimation to determine the distribution laws of air pollutants and thereby specify more effective measures to realize the high-quality development of the city. [ABSTRACT FROM AUTHOR]

• Filter, wrapper, embedded and ensemble feature selection approaches (FS) were evaluated. • The proposed ensemble FS incorporates ten individual selectors to identify top significant variables. • Ensemble FS produced the highest predictive variable subset for mapping SOM. • XGBoost outperformed RF in mapping SOM in subtropical restored forests. Mapping Soil organic matter (SOM) over a complex forest landscape is challenging due to the difficulty in selecting the most insightful variables from high-dimensional datasets in the recent explosion of geospatial-data. Feature selection (FS) is necessary to reduce data redundancy and noise as well as to achieve more reliable SOM spatial predictions. However, it is still unclear that which is most effective among various FS methods in mapping SOM. Therefore, four types of FS approaches (i.e., filter, wrapper, embedded and ensemble) were adopted to generate optimum variable subsets from an original variable dataset of 60 candidates, respectively, for mapping SOM of restored forest land in a typical subtropical region of southern China. The most used methods for each type of FS approaches were selected in this study, including three filters (Chi-square, InfoGain and pearson correlation analysis), three wrappers (genetic algorithm, simulated annealing algorithm and support vector machine-recursive feature elimination) and three embedded methods (Boruta, random forest (RF) and extreme gradient boosting (XGBoost)), as well as an ensemble method (robust rank aggreg algorithm (RRA)). Meanwhile, the RF and XGBoost models were applied with a 10-fold cross-validation method to compare the relative advantages of the different FS methods in SOM mapping, by utilizing the correlation coefficients R2 between observed and predicted values and predicting errors of root mean square error (RMSE). The results show that the SOM prediction accuracies with optimized variable subsets generated by the different FS methods are better than those with full variables, yet the improvements of prediction performance are different among the four types of FS approaches. The ensemble method (RRA) is superior to the other three types of approaches with an average RMSE reduction of 9.16% comparing to that without using FS methods, followed by wrapper and embedded methods which obtained the average RMSE reduction by 7.81%, 7.32%, respectively, and the filter methods are the weakest in the RMSE reduction with slight decreases of 4.32%. The XGBoost model achieved a better performance in predicting SOM than the RF model regardless of input variables, and the XGBoost model combined with RRA FS method shows the greatest potential to map SOM in the restored forest land. This study provides a reference for obtaining more parsimonious and robust variable sets from the available big geo-data freely for soil mapping in other areas. [ABSTRACT FROM AUTHOR]

Quantitatively mapping forest aboveground biomass (AGB) is of great significance for the study of terrestrial carbon storage and global carbon cycles, and remote sensing-based data are a valuable source of estimating forest AGB. In this study, we evaluated the potential of machine learning algorithms (MLAs) by integrating Gaofen-1 (GF1) images, Sentinel-1 (S1) images, and topographic data for AGB estimation in the Dabie Mountain region, China. Variables extracted from GF1 and S1 images and digital elevation model data from sample plots were used to explain the field AGB value variations. The prediction capability of stepwise multiple regression and three MLAs, i.e., support vector machine (SVM), random forest (RF), and backpropagation neural network were compared. The results showed that the RF model achieved the highest prediction accuracy (R2 = 0.70, RMSE = 16.26 t/ha), followed by the SVM model (R2 = 0.66, RMSE = 18.03 t/ha) for the testing datasets. Some variables extracted from the GF1 images (e.g., normalized differential vegetation index, band 1-blue, the mean texture feature of band 3-red with windows of 3 × 3), S1 images (e.g., vertical transmit-horizontal receive and vertical transmit-vertical receive backscatter coefficient), and altitude had strong correlations with field AGB values (p < 0.01). Among the explanatory variables in MLAs, variables extracted from GF1 made a greater contribution to estimating forest AGB than those derived from S1 images. These results indicate the potential of the RF model for evaluating forest AGB by combining GF1 and S1, and that it could provide a reference for biomass estimation using multi-source images. [ABSTRACT FROM AUTHOR]

Land use and land cover change (LUCC) modeling has continuously been a major research theme in the field of land system science, which interprets the causes and consequences of land use dynamics. In particular, models that can obtain long-term land use data with high precision are of great value in research on global environmental change and climate impact, as land use data are important model input parameters for evaluating the effect of human activity on nature. However, the accuracy of existing reconstruction and prediction models is inadequate. In this context, this study proposes an integrated convolutional neural network (CNN) LUCC reconstruction and prediction model (CLRPM), which meets the demand for fine-scale LUCC reconstruction and prediction. This model applies the deep learning method, which far exceeds the performance of traditional machine learning methods, and uses CNN to extract spatial features and provide greater proximity information. Taking Baicheng city in Northeast China as an example, we verify that CLRPM achieved high-precision annual LUCC reconstruction and prediction, with an overall accuracy rate 9.38% higher than that of the existing models. Additionally, the error rate was reduced by 49.5%. Moreover, this model can perform multilevel LUCC classification category reconstructions and predictions. This study casts light on LUCC models within the high-precision and fine-grained LUCC categories, which will aid LUCC analyses and help decision-makers better understand complex land-use systems and develop better land management strategies. [ABSTRACT FROM AUTHOR]

Many studies have attempted to predict soil organic matter (SOM), whereas mapping high-precision and high-resolution SOM maps remains a challenge due to the difficulty of selecting appropriate satellite data sources and prediction algorithms. This study aimed to investigate the influence of different remotely sensed images and machine learning algorithms on SOM prediction. We constructed two comparative experiments, i.e., full-band and common-band variable datasets of Sentinel-2A and MODIS images using Google Earth Engine (GEE). The predictive performances of random forest (RF), artificial neural network (ANN), and support vector regression (SVR) algorithms were evaluated, and the SOM map was generated for the Songnen Plain. Results showed that the model based on the full-band Sentinel-2A dataset achieved the best performance. The application of Sentinel-2A data resulted in mean relative improvements (RIs) of 7.67% and 5.87%, respectively. The RF achieved a lower root mean squared error (RMSE = 0.68%) and a higher coefficient of determination (R2 = 0.67) in all of the predicted scenarios than ANN and SVR. The resultant SOM map accurately characterized the SOM spatial distribution. Therefore, the Sentinel-2A data have obvious advantages over MODIS due to their higher spectral and spatial resolutions, and the combination of the RF algorithm and GEE is an effective approach to SOM mapping. [ABSTRACT FROM AUTHOR]

In this study, Logistics Regression (LR), Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting Machine (GBM), and Multilayer Perceptron (MLP) machine learning algorithms are combined with GIS techniques to map landslide susceptibility in Shexian County, China. By using satellite images and various topographic and geological maps, 16 landslide susceptibility factor maps of Shexian County were initially constructed. In total, 502 landslide and random safety points were then using the "Extract Multivalues To Points" tool in ArcGIS, parameters for the 16 factors were extracted and imported into models for the five algorithms, of which 70% of samples were used for training and 30% of samples were used for verification, which makes sense for date symmetry. The Shexian grid was converted into 260130 vector points and imported into the five models, and the natural breakpoint method was used to divide the grid into four levels: low, moderate, high, and very high. Finally, by using column results gained using Area Under Curve (AUC) analysis and a grid chart, susceptibility results for mapping landslide prediction in Shexian County was compared using the five methods. Results indicate that the ratio of landslide points of high or very high levels from LR, SVM, RF, GBM, and MLP was 1.52, 1.77, 1.95, 1.83, and 1.64, and the ratio of very high landslide points to grade area was 1.92, 2.20, 2.98, 2.62, and 2.14, respectively. The success rate of training samples for the five methods was 0.781, 0.824, 0.853, 0.828, and 0.811, and prediction accuracy was 0.772, 0.803, 0.821, 0.815, and 0.803, respectively; the order of accuracy of the five algorithms was RF > SVM > MLP > GBM > LR. Our results indicate that the five machine learning algorithms have good effect on landslide susceptibility evaluation in Shexian area, with Random Forest having the best effect. [ABSTRACT FROM AUTHOR]

Due to tremendous reception on digital learning platforms, many online users tend to register for online courses in MOOC offered by many prestigious universities all over the world and gain a lot on cutting edge technologies in niche courses. As the reception of online courses is increasing on one side, there have been huge dropouts of participants in the online courses causing serious problems for the course owners and other MOOC administrators. Hence, it is deemed necessary to find out the root causes of course dropouts and need to prepare a workable solution to prevent that outcome in the future. In this connection, the authors made use of three machine learning algorithms such as support vector machine, random forest, and conditional random fields. The huge samples of datasets were downloaded from the Open University of China, that is, almost 7K student profiles were extracted for the empirical analysis. The datasets were loaded into a confusion matrix and analyzed for the accuracy, precision, recall, and f-score of the model. [ABSTRACT FROM AUTHOR]

Predicting next-day heat load curves is essential to guarantee sufficient heat supply and optimal operation of district heat systems (DHSs). Existing studies have mainly investigated one-step ahead forecasting methods, which can predict a single value at a future time step. To predict heat load curves, multi-step ahead forecasting methods are needed. This study proposes a machine learning-based framework for multi-step ahead DHS heat load forecasting. Specifically, support vector regression, deep neural network, and extreme gradient boosting (XGBoost) are respectively used as the base learner to develop forecasting models. Two multi-step ahead forecasting methods, i.e. direct strategy and recursive strategy, adopt the learnt models to generate predictions. A DHS in China is used as the case study to comprehensively assess the performance of these two forecasting strategies. Recursive strategy using the XGBoost-based forecasting model can achieve the most accurate and stable predictions with a value of 10.52% for the coefficient of variation of root mean square error. Furthermore, the modeling process of recursive strategy is much more convenient than that of direct strategy. The research shows that the recursive strategy is a better solution to multi-step ahead forecasting than the direct strategy with respect to accuracy, prediction stability, and modeling process. • A framework for multi-step ahead heat load forecasting is proposed. • Direct and recursive strategies are used to predict daily heat load curves. • Applicability of direct and recursive strategies is assessed from three aspects. • Recursive strategy slightly outperforms direct strategy in accuracy and stability. • Modeling process of recursive strategy is simpler than that of direct strategy. [ABSTRACT FROM AUTHOR]