Search

China has resolved its overall regional poverty in 2020 by attaining moderate societal prosperity. The country has entered a new development stage designed to achieve its second centenary goal. However, ecological fragility and risk susceptibility have increased the risk of returning to ecological poverty. In this paper, the Liupan Mountain Region of China was used as a case study, and the counties were used as the scale to reveal the spatiotempora differentiation and influcing factors of the risk of returning to poverty in study area. The indicator data for returning to ecological poverty from 2011–2020 were collected and summarized in three dimensions: ecological, economic and social. The autoregressive integrated moving average model (ARIMA) time series and exponential smoothing method (ES) were used to predict the multidimensional indicators of returning to ecological poverty for 61 counties (districts) in the Liupan Mountain Region for 2021–2030. The back propagation neural network (BPNN) and geographic information system (GIS) were used to generate the spatial distribution and time variation for the index of the risk of returning to ecological poverty (RREP index). The results show that 1) ecological factors were the main factors in the risk of returning to ecological poverty in Liupan Mountain Region. 2) The RREP index for the 61 counties (districts) exhibited a downward trend from 2021–2030. The RREP index declined more in medium- and high-risk areas than in low-risk areas. From 2021 to 2025, the RREP index exhibited a slight downward trend. From 2026 to 2030, the RREP index was expected to decline faster, especially from 2029–2030. 3) Based on the RREP index, it can be roughly divided into three types, namely, the high-risk areas, the medium-risk areas, and the low-risk areas. The natural resource conditions in low-risk areas of returning to ecological poverty, were better than those in medium- and high-risk areas. [ABSTRACT FROM AUTHOR]

Food security is a vital material foundation for a nation's development and has been a topic of significant concern on the international stage in recent years. With a population exceeding 1.4 billion, China is not only a major producer but also a substantial consumer of food. Ensuring food security in China is not only a top priority for its socio-economic development but also a driving force in maintaining the stability of the global food supply chain and reducing the number of hungry people worldwide. However, a lack of comprehensive research into the Chinese food security system remains. This study addresses this gap by constructing a comprehensive evaluation framework encompassing four dimensions: food supply, accessibility, production stability, and sustainability. Utilizing the Moran's Index and generating LISA (Local Indicators of Spatial Association) maps, we analyze the spatial correlations of food security. The Dagum Gini coefficient and kernel density estimation are applied to assess heterogeneity and spatial disparities. Furthermore, this research employs the Exponential Smoothing (ETS) model to forecast food security trends. The findings reveal that the overall composite food security score exhibited fluctuations, initially increasing and reaching its peak of 0.407 in 2003, followed by a subsequent sharp decline after 2019. Spatially, food security exhibits correlations, with the Huang-Huai-Hai Plain and Northeast regions consistently showing high-high clustering. In contrast, the Western and Southern regions exhibit low-low clustering at specific periods. The Dagum Gini coefficient indicates that overall food security disparities are relatively small. However, these disparities have gradually expanded in recent years, with inter-group differences becoming predominant after 2005. As indicated by the kernel density estimation, the dynamic distribution of food security initially widens and then narrows, suggesting a shift from dispersed to concentrated data distribution. This phenomenon is accompanied by polarization and convergence trends, particularly evident after 2015. According to the ETS model, the study forecasts a substantial risk of declining food security in China over the next decade, largely influenced by the ongoing pandemic. In conclusion, this research provides a comprehensive assessment of the changing status of food security in China. It offers early warnings through predictive analysis, addressing the existing research gaps in the field of food security. [ABSTRACT FROM AUTHOR]

Under the long-term implementation of the low fare policy for public transport in China, the phenomenon of unsustainable subsidies is increasingly obvious. We take into account three related stakeholders including the public, operators and government, and comprehensively consider various factors affecting public transport subsidies, and establish a linkage mechanism of the subsidy based on passenger ridership, cost, fare and service quality. We propose a combination forecasting model (CFM) composed of quadratic exponential smoothing model (QESM) and Grey theory model (GM) to forecast passenger ridership. The results show that the CFM improves the forecast accuracy of passenger ridership, the factors considered in the linkage mechanism are more comprehensive, and the results are reasonable. By using this linkage mechanism, the government or operators can establish long-term or mid-term development plans to provide a systematic and theoretical reference, which is conducive to the healthy and stable development of public transport. [ABSTRACT FROM AUTHOR]

Water ecological carrying capacity (WECC) is a crucial index for measuring regional sustainable development. To investigate the evolution of WECC in the Yangtze River Economic Belt (YREB), this study constructed a comprehensive index system consisting of 23 indicators from six interconnected dimensions of water systems. The back propagation neural network (BPNN) model was used to quantify WECC in YREB from 2010 to 2021, and ArcGIS was utilized to visualize the distribution of WECC. To identify sensitive indicators under subsystems, sensitivity analysis was conducted based on the one-at-a-time (OAT) method. Additionally, time-series prediction of WECC was performed using the exponential smoothing (ES) method. Finally, the coupling coordinated degree (CCD) of subsystems in each province from 2010 to 2021 was calculated. The results indicated that the average WECC in YREB gradually increased from 2010 to 2021, with significant provincial differences. Sensitivity analysis revealed that R1, U2, Q4, S2, M3 and B1 had the most substantial impacts on the WECC of subsystems (Sub-CC). The fitting curve between the CCD and WECC showed that as CCD increased, the growth rate of WECC gradually slowed down. Based on these findings, relevant suggestions were proposed to improve WECC and promote the regional sustainable development in YREB. [ABSTRACT FROM AUTHOR]

The prevalence of cybercrime has emerged as a critical issue in contemporary society because of its far-reaching financial, social, and psychological implications. The negative effects of cyber-attacks extend beyond financial losses and disrupt people's lives on social and psychological levels. Conventional practice involves cyber experts sourcing data from various outlets and applying personal discernment and rational inference to manually formulate cyber intelligence specific to a country. This traditional approach introduces personal bias towards the country-level cyber reports. However, this paper reports a novel approach where country-level cyber intelligence is automatically generated with artificial intelligence (AI), employing cyber-related social media posts and open-source cyber-attack statistics. Our innovative cyber threat intelligence solution examined 37,386 tweets from 30,706 users in 54 languages using sentiment analysis, translation, term frequency–inverse document frequency (TF-IDF), latent Dirichlet allocation (LDA), N-gram, and Porter stemming. Moreover, the presented study utilized 238,220 open-intelligence cyber-attack statistics from eight different web links, to create a historical cyber-attack dataset. Subsequently, AI-based algorithms, like convolutional neural network (CNN), and exponential smoothing were used for AI-driven insights. With the confluence of the voluminous Twitter-derived data and the array of open-intelligence cyber-attack statistics, orchestrated by the AI-driven algorithms, the presented approach generated seven-dimensional cyber intelligence for Australia and China in complete automation. Finally, the topic analysis on the cyber-related social media messages revealed seven main themes for both Australia and China. This methodology possesses the inherent capability to effortlessly engender cyber intelligence for any country, employing an autonomous modality within the realm of pervasive computational platforms. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Chongqing University of Technology (Natural Science) is the property of Chongqing University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this paper, we propose a structured multiple-index model (SMIM) for ultrahigh-dimensional data analysis. The proposed model takes many commonly used semiparametric models as special cases, including the stochastic frontier model, single-index model, and additive-index model. We estimate all of the functions and parameters based on a full likelihood-type function. As a result, the proposed estimators are shown to be semiparametrically efficient, consistent in terms of selection and estimation, and asymptotically normal. The computation is challenging owing to the combination of nonconvexity of the likelihood function, the nonsmoothness of the penalty term, and the large number of functions. To solve the computational problem, we blend spline and kernel smoothing with a majorized coordinate descendent algorithm, making the implementation easy to perform using existing packages. Intensive simulation studies show that the proposed estimation procedure outperforms alternatives for various cases. Finally, we apply the proposed SMIM and estimation procedure to a real data set from one of China’s largest liquor companies, successfully identifying the 31, from 2051, most important factors affecting the sale of liquor. [ABSTRACT FROM AUTHOR]

China has implemented a series of long-term measures to control the spread of COVID-19, however, the effects of these measures on other chronic and acute respiratory infectious diseases remain unclear. Tuberculosis (TB) and scarlet fever (SF) serve as representatives of chronic and acute respiratory infectious diseases, respectively. In China's Guizhou province, an area with a high prevalence of TB and SF, approximately 40,000 TB cases and hundreds of SF cases are reported annually. To assess the impact of COVID-19 prevention and control on TB and SF in Guizhou, the exponential smoothing method was employed to establish a prediction model for analyzing the influence of COVID-19 prevention and control on the number of TB and SF cases. Additionally, spatial aggregation analysis was utilized to describe spatial changes in TB and SF before and after the COVID-19 outbreak. The parameters of the TB and SF prediction models are R2 = 0.856, BIC = 10.972 and R2 = 0.714, BIC = 5.325, respectively. TB and SF cases declined rapidly at the onset of COVID-19 prevention and control measures, with the number of SF cases decreasing for about 3–6 months and the number of TB cases remaining in decline for 7 months after the 11th month. The spatial aggregation of TB and SF did not change significantly before and after the COVID-19 outbreak but exhibited a marked decrease. These findings suggest that China's COVID-19 prevention and control measures also reduced the prevalence of TB and SF in Guizhou. These measures may have a long-term positive impact on TB, but a short-term effect on SF. Areas with high TB prevalence may continue to experience a decline due to the implementation of COVID-19 preventive measures in the future. [ABSTRACT FROM AUTHOR]

Flight demand forecasting is a particularly critical component for airline revenue management because of the direct influence on the booking limits that determine airline profits. The traditional flight demand forecasting models generally only take day of the week (DOW) and the current data collection point (DCP) adds up bookings as the model input and uses linear regression, exponential smoothing, pick-up as well as other models to predict the final bookings of flights. These models can be regarded as time series flight demand forecasting models based on the interval between the current date and departure date. They fail to consider the early bookings change features in the specific flight pre-sale period, and have weak generalization ability, at last, they will lead to poor adaptability to the random changes of flight bookings. The support vector regression (SVR) model, which is derived from machine learning, has strong adaptability to nonlinear random changes of data and can adaptively learn the random disturbances of flight bookings. In this paper, flight bookings are automatically divided into peak, medium, and off (PMO) according to the season attribute. The SVR model is trained by using the vector composed of historical flight bookings and adding up bookings of DCP in the early stage of the flight pre-sale period. Compared with the traditional models, the priori information of flight is increased. We collect 2 years of domestic route bookings data of an airline in China before COVID-19 as the training and testing datasets, and divide these data into three categories: tourism, business, and general, the numerical results show that the SVR model significantly improves the forecasting accuracy and reduces RMSE compared with the traditional models. Therefore, this study provides a better choice for flight demand forecasting. [ABSTRACT FROM AUTHOR]

Cross-border e-commerce has gradually emerged with the popularity of the Internet in the world. It represents a new business model of e-commerce. The public is still the same as the daily shopping model on e-commerce platforms. However, cross-border e-commerce can be divided into two parts at home and abroad. With the continuous emergence of new social models such as the Metaverse, the data generated by online social networking is becoming larger and larger, and data science has developed accordingly, and the effective use of data has become the key to the rapid development of various industries. Big data technology is the key to data science. A data processing and analysis technology generated in the development process, its use in the development of cross-border e-commerce enterprises will help enterprises to create products that are more in line with the actual requirements of the masses, increase the profits of enterprise operations, and allow enterprises to complete the era of data science smooth transition. This paper first describes the big data technology and related theories of cross-border e-commerce, and then takes Alibaba as an example to describe the SWOT analysis of my country's cross-border e-commerce enterprises, and then analyzes the development dilemma of China's cross-border e-commerce enterprises, and finally designs The development strategy of my country's cross-border e-commerce enterprises based on big data is presented. Through a questionnaire survey of cross-border e-commerce enterprises that use big data technology in my country, it is found that cross-border e-commerce enterprises that combine big data technology have a greater understanding of overseas customers' consumption psychology, integration of warehousing and logistics resources, and talent training. promote. [ABSTRACT FROM AUTHOR]

Copyright of Railway Investigation & Surveying is the property of Railway Investigation & Surveying Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

To reasonably ensure good groundwater quality for the safety of users, it is necessary to monitor the evolution of physicochemical parameters of water. This study assessed the evolution of groundwater quality in Xianyang City (Central China) for 6 years and forecasted the water quality index (WQI) using the automatic exponential smoothing model (AESM). The results revealed that pH, Mn, NH4+, HCO3− and Ca2+ increased significantly within 6 years. However, there was a slight decrease in K+, NO3−, SiO2, and Mg2+. Over the course of 6 years, seven out of fourteen wells maintained excellent or good water quality, whereas groundwater quality in the remaining wells fluctuated from excellent to good or from good to poor. The WQI index increased for nine wells, and five of them including well 173, well 457, well 46, well 26, and well C4 manifested an exponential increase of WQI, which was mainly caused by high concentrations of Mn, NH4+, and NO3−. The forecast accuracy verified by MASE and sMAPE, which ranged from 0.35 to 1.06 and from 0.04 to 0.69, respectively, demonstrated that the predictions were reliable. The AESM was tested in groundwater quality prediction, indicating that it might be used for actual scientific studies. [ABSTRACT FROM AUTHOR]

Landslide disasters cause serious property losses and casualties every year. Landslide displacement prediction is fundamental for mitigating landslide disasters. Several approaches have been used to predict landslide displacement, yet a more accurate and reliable displacement prediction still has a poor understanding of landslide early warning systems for landslide mitigation, due to limited data and mutational displacements. To boost the robustness and accuracy of landslide displacement prediction, this paper assembled a new hybrid model containing the local mean decomposition (LMD), innovations state space models for exponential smoothing (ETS), and the temporal convolutional network (TCN). The proposed model, which is based on over 10 years of long-term time series monitoring GPS data, was tested on the selected case—stepwise Baijiabao landslide in the Three Gorges Reservoir area of China (TGRA) was tested by the proposed model. The results presented that the LMD–ETS–TCN model has the best performance in comparison with other benchmark models. Compared with autoregressive integrated moving average (ARIMA), support vector regression (SVR), and long short-term memory neural network (LSTM), the accuracy was noticeably improved by an average of 40.9%, 46.2%, and 22.1%, respectively. The robustness and effectiveness of the presented approach are attested, and it has discernible improvements for landslide displacement prediction. [ABSTRACT FROM AUTHOR]

It is of great significance to evaluate the seawater intrusion degree and predict the change of water quality for coastal groundwater resources. This study takes Dagu River in Jiaodong Peninsula of North China as the target area and combines the relevant theoretical research results to build a seawater intrusion fuzzy analytic hierarchy process (AHP) evaluation model. Five sensitive indicators of water quality, such as Cl−, SO42−, NO3−, TH, and TDS, were selected to evaluate the seawater intrusion level of the long series monitoring data in Xilaiwan, Guanzhuang, and Ligezhuang of Dagu River Basin by using the basic fuzzy mathematics principles and the improved hierarchical analysis method. In this study, the cubic exponential smoothing method was applied to predict groundwater quality change in Dagu River Basin. In order to evaluate the change of seawater intrusion in detail and make timely prediction, this paper innovatively divided the classification standard of seawater intrusion degree based on relevant norms and scholars' research and predicted the evaluation level of seawater intrusion by using long series historical observation data combined with fuzzy analytic hierarchy process. The cubic exponential smoothing method which has the characteristics of simple and fast was introduced to fit the observation elements, and the historical data were used to verify the prediction of the future development trend. Compared with the evaluation results of seawater intrusion by traditional methods, this study can reflect the whole development trend of seawater intrusion in detail and has the characteristics of more reasonable, accurate, and practical. It also provides a certain reference for the future seawater intrusion prevention. In addition to this case, the method proposed in this study will be applicable to a wider range of coastal zones, providing a new idea for the rational management and control of coastal groundwater resources. [ABSTRACT FROM AUTHOR]

The ecological compensation mechanism is regarded as the direction for the future management of the ecological environment of the river basin, which has become a global concern. Ex-post assessments on the performance of ecological compensation programs contribute to further improvement and optimization in the process of exploration. This study proposes a novel performance assessment approach to address the issue of environmental efficiency evaluation with uncertainty by systematically integrating data envelopment analysis (DEA), bootstrap, regression, and exponential smoothing. The last two methods are used to fill in missing data, DEA super-SBM is applied to measure the performance, and bootstrap is adopted to stimulate more data. This approach is applied to performance measurement of Xin'an river basin in China. Validated by benchmark comparisons and statistical tests, the outcomes indicate that this integrated conceptual method can serve as an effective way to gauge environmental efficiency with ecological compensation when missing data are presented. The results obtained from the case highlight the limited positive effect of ecological compensation. Marginal utility brought about by the ecological compensation investment fund is declining, and the fund utilization in some projects also appears to be low. Consequently, suggestions for future optimization on fund and performance management, evaluation method and compensation mode are provided. [ABSTRACT FROM AUTHOR]

Objective Variation in crop yield is the consequence of many natural and anthropogenic factors, and disentangling their impacts is important for improving agricultural management but difficult. The purpose of this paper is to propose and compared different methods to isolate the impacts of meteorological change on crop yield based on long time series of maize yield in Yellow River-water irrigated region in Ninxia province of China. Method The analysis is based on maize yield measured from1988 to 2019 in 6 counties located in the Yellow River-watered irrigation areas. We compared three methods for the separation: five -year moving average method, quadratic exponential smoothing method, and five-point quadratic smoothing method. The consistent correlation coefficient, trend coincidence conformity analysis method, consistency of climate change characteristics, which lead to the same rise-fall in meteorological yield, were used as the evaluation criteria. Their applicability and rationality were compared and analyzed. All methods were calibrated based on the relationship between meteorological factors and maize yield. Result All methods can fit the yield trend well. Compared with the average yield trend, the consistency correlation coefficients of all three methods were >0.5, suggesting that there was no significant difference between these methods for fitting the yield trend. The advantage of the quadratic exponential smoothing method and the five-point quadratic smoothing method is that they accurately describe the change in the yield as affected by national productivity and national policy. The change in the yield due to meteorological factors estimated by the five-point quadratic smoothing method described the effect of inter-annual meteorological factors better, and its associated meteorological yield model is able to describe the relationship between the meteorological factors and the maize yield. Conclusion omprehensive analysis showed that the five-point quadratic smoothing method modeled the yield change due to meteorological factors better than the other two methods. [ABSTRACT FROM AUTHOR]

Defense Meteorological Satellite Program Operational Linescan System (DMSP-OLS) and Suomi National Polar-orbiting Partnership Visible Infrared Imaging Radiometer Suite (SNPP-VIIRS) data are valuable records of nighttime lights (NTLs) in analyzing socioeconomic development. However, inconsistencies between these data have severely restricted long time-series analyses. Published time-series NTL data sets are not widely available or accurate because the DMSP-OLS calibration is inadequate and some missing data in the SNPP-VIIRS data are seldom considered for patching. To address these issues, we calibrated DMSP-OLS data (1992–2013) by using a quadratic model based on a “pseudo-invariant pixel” method. Thereafter, an exponential smoothing model was used to predict and patch missing data in the monthly SNPP-VIIRS data (2013–2019). Outliers and noise were also removed from the annual data. In addition, a sigmoid model was employed to generate improved simulated DMSP-OLS (SDMSP-OLS) data (2013–2019), which were appended with the calibrated DMSP-OLS data (1992–2013) to develop improved DMSP-OLS-like data (1992–2019) in China. Finally, we qualitatively and quantitatively compared these data with published NTL data to examine data availability. Results showed that choosing invariant pixels to calibrate DMSP-OLS data can minimize discontinuity. The correlation between the SNPP-VIIRS data synthesized by the patched monthly SNPP-VIIRS data and the official annual SNPP-VIIRS data in 2015 ($R^{2} =0.931$) and 2016 ($R^{2} =0.930$) was higher than those of the two existing correction methods with $R^{2}$ values below 0.90. Spatial patterns of pixels in the improved SDMSP-OLS data in 2013 were more similar with the DMSP-OLS data than those in the published data. Strong correlations likewise existed between the total (average) pixel values of the improved SDMSP-OLS data (2013–2019) and the DMSP-OLS data in 2012. We also found that the improved DMSP-OLS-like data held strong linear correlations with different statistics, the average $R^{2}$ values of which were 0.931 and 0.654 at the national and provincial levels, respectively. Meanwhile, the average regression $R^{2}$ values between the two published data sets and statistics were 0.858/0.506 and 0.911/0.611, respectively. Our study has proven that the improved DMSP-OLS-like data (1992–2019) have immense potential to effectively evaluate socioeconomic development and anthropic activities. [ABSTRACT FROM AUTHOR]

Objectives: The objective of this study is to compare the various nonlinear and time series models in describing the course of the coronavirus disease 2019 (COVID-19) outbreak in China. To this aim, we focus on 2 indicators: the number of total cases diagnosed with the disease, and the death toll. Methods: The data used for this study are based on the reports of China between January 22 and June 18, 2020. We used nonlinear growth curves and some time series models for prediction of the number of total cases and total deaths. The determination coefficient (R2), mean square error (MSE), and Bayesian Information Criterion (BIC) were used to select the best model. Results: Our results show that while the Sloboda and ARIMA (0,2,1) models are the most convenient models that elucidate the cumulative number of cases; the Lundqvist-Korf model and Holt linear trend exponential smoothing model are the most suitable models for analyzing the cumulative number of deaths. Our time series models forecast that on 19 July, the number of total cases and total deaths will be 85,589 and 4639, respectively. Conclusion: The results of this study will be of great importance when it comes to modeling outbreak indicators for other countries. This information will enable governments to implement suitable measures for subsequent similar situations. [ABSTRACT FROM AUTHOR]

Analyzing the drivers of energy demand and predicting energy consumption can help to shape national policies on energy transformation and energy security. This paper estimates the demand for the various forms of energy consumed in China. A least absolute shrinkage and selection operator-random forest (Lasso-RF) two-stage model is first proposed to identify the drivers of the demand in energy. A support vector regression-compositional data second exponential smoothing (SVR-CDSES) model is then established to forecast the demand for coal, oil, natural gas, and primary electricity. The empirical analysis results inform that China's appetite for energy in the next decade will grow at an annualised rate of 2.68%. The demand for coal will rise initially and then fall at a rate of 0.79% per annum. The annual growth rate in the demand for oil will be relatively stable at 3.52%. However, the demand for natural gas and primary electricity will increase rapidly, with the highest annual growth rate at 8.05%, pointing to greater awareness of clean energy. • Study the drivers of China's energy demand and the trend of energy demand. • A new two-stage model is presented to identify the drivers in energy demand. • A new compound Support Vector Regression model is proposed to predict energy demand. • Some managerial insights and policy guidelines for energy demand are provided. [ABSTRACT FROM AUTHOR]

• Yield monitoring system design. • Yield reconstruction model. • Analysis of spatiotemporal variability. An impact-type grain yield monitoring system was developed and tested in laboratory and field to meet the requirement of precision agriculture in China. The system consisted of a mass flow sensor, a clean elevator speed sensor, a ground speed sensor, a moisture content sensor, a controller area network-bus processor, a global navigation satellite system receiver, and an industrial personal computer. The system was used to collect yield data and create the grid map in a field of winter wheat from 2018 to 2019. Laboratory and field calibrations were implemented to verify the accuracy of the mass flow sensor and the moisture content sensor. The errors of time delay and parking were removed, and exponential smoothing was performed in the data preprocessing. A yield reconstruction method was proposed to calculate the yield data of the sampling points based on the empirical model between flow voltage and grain weight. The predicted errors of total yield in both years were less than 5%, which indicated that the accuracy of yield monitoring system was high. An averaging method was used to generate the yield grid map. The spatiotemporal variation of the two-year yield was analyzed. Hierarchical management for precision farming was studied according to spatial variability ( CV S = 31%) in 2018. Results showed that the temporal variability was CV T = 34%, which confirmed that the yield increased by about 0.3 t compared with the previous year. Moreover, the variability of the grid map showed the distribution of the yield reduction area, which was related to the characteristics of the seeds and the regional properties of the soil. The results proved that the developed yield monitoring system can be used as a tool to provide a reference for precision agriculture and high-yield management of winter wheat. [ABSTRACT FROM AUTHOR]

This paper compares the performance of five popular machine learning methods, namely, particle swarm optimization–extreme learning machine (PSO–ELM), particle swarm optimization–kernel extreme learning machine (PSO–KELM), particle swarm optimization–support vector machine (PSO–SVM), particle swarm optimization–least squares support vector machine (PSO–LSSVM), and long short-term memory neural network (LSTM), in the prediction of reservoir landslide displacement. The Baishuihe, Shuping, and Baijiabao landslides in the Three Gorges reservoir area of China were used for case studies. Cumulative displacement was decomposed into trend displacement and periodic displacement by the Hodrick–Prescott filter. The double exponential smoothing method and the five machine learning methods were used to predict the trend and periodic displacement, respectively. The five machine learning methods are compared in three aspects: highest single prediction accuracy, mean prediction accuracy, and prediction stability. The results show that no method performed the best for all three aspects in the three landslide cases. LSTM and PSO–ELM achieved better single prediction accuracy, but worse mean prediction accuracy and stability. PSO–KELM, PSO–LSSVM, and PSO–SVM always yielded consistent predictions with slight variations. On the whole, PSO–KELM and PSO–LSSVM are recommended for their superior mean prediction accuracy and prediction stability. • The performances of machine learning methods depend on the selected evaluation index and reservoir landslide case. • Using single prediction accuracy for evaluating the superiority of machine learning methods may be unreliable. • PSO-KELM and PSO-LSSVM are recommended for their superior mean prediction accuracy and prediction stability. [ABSTRACT FROM AUTHOR]