Search

The total electron content (TEC) is a crucial parameter for ionosphere monitoring, and the development of accurate TEC estimation and prediction models is significant for high-precision positioning. However, establishing a high-precision ionospheric model that can effectively improve prediction accuracy remains a challenging task. We propose a method that combines random forest with a Bi-LSTM neural network to establish a high-precision ionospheric prediction model. The random forest algorithm is used for regression analysis and to estimate the variable importance of input parameters. We use observation data from the Crustal Movement Observation Network of China and International GNSS Services while estimating the relative importance of 14 input variables at different latitudes. The results demonstrate that our proposed model achieves more efficient and accurate predictions, with a correlation coefficient of 0.96, indicating significant improvement in accuracy compared to traditional RNN and LSTM models. Overall, the Bi-LSTM forecast model based on random forest can effectively capture the temporal and spatial variation characteristics of ionospheric TEC in China. This enables the model to provide accurate ionospheric information for positioning users, thereby improving the precision of positioning applications. [ABSTRACT FROM AUTHOR]

• Ensemble models are explored for prediction of Solar radiation and PV power. • Meteorological variables affecting solar radiation and PV power are analyzed. • An ensemble of Random Forest and CatBoost algorithm is developed. • Improved prediction accuracy is shown using Bayesian Model Averaging. Although solar energy is renewable, environmental conditions can make it unpredictable, which makes it difficult to maintain a consistent supply of electricity. Accurate solar forecast techniques are essential to overcome this. Therefore, an efficient model for predicting solar irradiance and PV power output is presented in this study. The suggested model is especially designed for Agartala city in Tripura, India, where the installed capacity of Grid Interactive Solar Power is extremely low. The proposed model for solar irradiance prediction involves utilizing a combination of the Random Forest and CatBoost algorithms. The model is developed using 10 years of solar data and other relevant meteorological parameters with a 1-hour interval and assessed at various stages to make long-term monthly predictions. In addition, we compared this model to other existing models for solar irradiance prediction. The study showcased the efficacy of Random Forest and CatBoost algorithms when utilized individually and as an ensemble. The results demonstrate the effectiveness of the Random Forest and CatBoost ensemble, with an impressive accuracy improvement of 6% and an R 2 -score of 86%. Additionally, in terms of RMSE and MAE, the model demonstrates superior performance with lower values of 83.8466 and 45.4011, respectively confirms to the proposed model's viability and suitability for long-term solar radiation and PV power predictions. [ABSTRACT FROM AUTHOR]

A robust short-term significant wave height (Hs) modelling framework based on an ensemble local mean decomposition method integrated with random forest (i.e. , En-RLMD-RF) is developed. The robust local mean decomposition (RLMD) decomposed the Hs data series into three subseries; amplitude modulation, frequency modulation and the low-frequency product function (PFs). The partial autocorrelation function was employed to determine the correlation-based significant predictor signals between the PFs at t 0 and t 1. Then the statistically significant PFs were incorporated into the random forest (RF) to construct the RLMD-RF model. The RLMD-RF based forecasted PFs were used again in the RF model as input predictors resulting in an ensemble-based RLMD-RF (i.e. , En-RLMD-RF) model for forecasting short-term Hs. The En-RLMD-RF model is validated and compared with RF, extreme learning machine (ELM) and multiple linear regression (MLR) models and their hybrids RLMD-RF, RLMD-ELM, RLMD-MLR, En-RLMD-ELM and En-RLMD-MLR counterparts using a set of performance metrics. The results demonstrated that the En-RLMD-RF model generates better forecasting accuracy against the benchmarking models. This study is beneficial for the application and optimization of more clean energy resources worldwide for sustained energy generation. [ABSTRACT FROM AUTHOR]

Multi-source data is widely used in the field of energy future prices forecasting, the improvement of forecast ability and data screening are becoming the focus of current research. In this paper, two tree-based models (namely, Random Forest and XGBoost model) are employed to predict China's crude oil future prices. The empirical analysis confirms that Random Forest and XGBoost model have superior prediction performances than benchmark and the XGBoost model performs best. An important finding is that there is a time gap between investor information search and processing because the prediction performance within the time lags is obviously superior than that of the current period. [ABSTRACT FROM AUTHOR]

• A spatiotemporal risk prediction of wildlife-vehicle collisions (WVCs) is presented. • A Bavarian dataset of around 731,000 accidents over a period of 10 years (2010–2019) was used. • Heterogenous datasets of impact factors are pre-processed using GIS. • Different machine learning (ML) approaches are applied - random forest, feedforward neural networks, and support vector machine classifier to predict the risk of WVCs. • ML algorithms provide high risk prediction quality of over 86% (double-weighted sensitivity and single-weighted specificity rate). • Spatial transferability and temporal forecast of WVC prediction was verified. Introduction: The technology in the automotive industry is becoming increasingly safer in the age of automated driving, but the number of accidents is still high, especially in wildlife-vehicle collisions (WVCs). To better avoid these accidents, patterns involved in these accidents must be detected. Method : This paper presents a spatiotemporal risk prediction of WVCs, including various road and environmental characteristics. A process of data preparation using GIS automated by Python scripts was developed to enable a spatiotemporal link of diverse features for the subsequent predictive data analysis. Different machine learning (ML) approaches were applied- random forest (RF), feedforward neural networks (FNN), and support vector machine classifier (SVM) - including automated ML to predict the risk of WVCs. Therefore, a dataset of approximately 731,000 accidents reported to the police in Bavaria over a period of 10 years (2010–2019) was used. In addition, non-accidents were randomly generated in Python over time and space for the supervised ML processes. As the actual risk probability for WVCs and non-WVCs is not entirely known, the impact of different training ratios between accidents and non-accidents was tested on the risk prediction quality (RPQ) (25%, 50%, 75%, 90% WVCs) of the double-weighted sensitivity and single-weighted specificity rate. Results: The test yielded high mean values of RPQ as an indicator for a suitable WVC prediction. Both RF (86.6%) and FNN (86.7%) were identified as suitable choices for WVC risk prediction in terms of RPQ. The SVM yielded a lower prediction quality, even though acceptable results could be achieved within a shorter runtime. Practical Applications: Spatial transferability was verified since the algorithm was trained on the dataset of Bavaria (excluding Upper Bavaria) and successfully tested in Upper Bavaria. WVC forecasts were also proven through training with datasets from 2010-2017 and in prediction for 2018 and 2019. [ABSTRACT FROM AUTHOR]

• This study provides inspiration to use breeding values for predicting variety yield. • Three machine learning algorithm approaches were used to develop prediction models. • The proposed model based on random forest achieved the highest prediction accuracy. • The model can provide support for selecting good varieties effectively in a district or a county. This study proposed a model that utilized machine learning algorithms to predict the yield of maize (Zea mays L.) cultivars. This will enable the selection of good cultivars with high yields that are suitable for planting in specific areas, such as a district or county. The breeding value of the cultivars and 11 types of time-series meteorological variables were selected as the input parameters of the model. The yield was set as the output parameter of the model. Random forest (RF), Levenberg–Marquardt neural network, and multilayer perceptron neural network algorithms were used to construct the model. The results showed that the RF outperformed the other algorithms in predicting the yield of maize cultivars by achieving the maximum coefficient of determination (R2) of 0.77 and minimal root-mean-square error of 320.25 kg/acre, mean absolute error of 229.84 kg/acre, and mean absolute percentage error of 7.1%. The constructed model can be used to effectively predict the yield of specific varieties to enable the selection of good varieties in specific areas, such as a district or county. A prediction of the yield of a specific maize cultivar in a particular planting environment can have considerable value. It can facilitate the objective identification of better adapting cultivars among farmers and support the precise promotion and planting of cultivars. [ABSTRACT FROM AUTHOR]

Forecasting concentrations of PM 2.5 is important due to its known impacts on public health and environment. However, PM 2.5 concentrations can vary significantly over short distances and time, which can be influenced by local emissions and short-term weather patterns. This spatiotemporal variability makes accurate PM 2.5 forecasting an inherently complex and challenging task. This study presented novel methodologies for short-term PM 2.5 concentration forecast by combining the atmospheric chemistry transport model Community Multiscale Air Quality Modeling System (CMAQ) with data-driven machine learning methods, namely long short-term memory (LSTM) and random forest (RF) models. The combined model system forecast PM 2.5 with 1 h, 1km × 1 km spatiotemporal resolution. The LSTM system forecast time-dependent PM 2.5 concentrations at observation sites with a maximum root mean square error (RMSE) of 3.66 μg/m3 for 1-hr forecast and 23.75 μg/m3 for 72-hr forecast, leveraging results obtained from the atmospheric transport model with RMSE of 45.81 μg/m3. Wavelet transform in the LSTM system allowed learning and prediction of PM 2.5 concentrations at different frequencies, capturing temporal variability of PM 2.5 at various time scales. The RF model predicted distributions of PM 2.5 concentrations by learning LSTM results and integrating crucial features such as CMAQ results, meteorological and topographical information. The feature significance of CMAQ results was the highest among the input features in RF models. Overall, the hybrid model could help with managing and mitigating the adverse effects of air pollution by enabling informed decision-making at the individual, community and policy levels. [Display omitted] • Long Short-Term Memory networks accurately forecast PM 2.5 concentrations at observation sites. • Spatial resolution of predicted PM 2.5 concentrations was enhanced using a random forest approach. • Forecasting decomposed sub-sequences of observed PM 2. 5 concentrations improved prediction accuracy. [ABSTRACT FROM AUTHOR]

Introduction: Although several risk stratification models have been developed to predict hip fracture mortality, efforts are still being placed in this area. Our aim is to (1) construct a risk prediction model for long-term mortality after hip fracture utilizing the RSF method and (2) to evaluate the changing effects over time of individual pre- and post-treatment variables on predicting mortality.Methods: 1330 hip fracture surgical patients were included. Forty-five admission and in-hospital variables were analyzed as potential predictors of all-cause mortality. A random survival forest (RSF) algorithm was applied in predictors identification. Cox regression models were then constructed. Sensitivity analyses and internal validation were performed to assess the performance of each model. C statistics were calculated and model calibrations were further assessed.Results: Our machine-learning RSF algorithm achieved a c statistic of 0.83 for 30-day prediction and 0.75 for 1-year mortality. Additionally, a COX model was also constructed by using the variables selected by RSF, c statistics were shown as 0.75 and 0.72 when applying in 2-year and 4-year mortality prediction. The presence of post-operative complications remained as the strongest risk factor for both short- and long-term mortality. Variables including fracture location, high serum creatinine, age, hypertension, anemia, ASA, hypoproteinemia, abnormal BUN, and RDW became more important as the length of follow-up increased.Conclusion: The RSF machine-learning algorithm represents a novel approach to identify important risk factors and a risk stratification models for patients undergoing hip fracture surgery is built through this approach to identify those at high risk of long-term mortality. [ABSTRACT FROM AUTHOR]

• Performance assessment and production estimation of grid-connected PV station. • Strong relationship between weather variables and the output of the PV station. • Strong interdependence of the meteorological parameters and performance parameters. • The distinction between the different variable correlation results is a hard task. • Feature selection and PCA analysis were used as dimensional reduction techniques. • Dimensionality reduced data was used for training random forest models. Meteorological variables have an important effect on the performance of a grid-connected photovoltaic station, in this paper, the impact of meteorological variables on the 6 mWp grid-connected photovoltaic station in the desert of Adrar has been explored through performance assessment and output power forecasting. The impact of the meteorological variables on performance parameters has been investigated by performing interdependence and correlation studies. A clear interdependence between some variables has been observed, but the complete separation between each variable correlation effect has proven to be a difficult task. A combination of random forest method and pre-processing techniques namely feature selection and Principal component analysis has been developed in order to predict power production using meteorological variables as inputs. The forecasting models have been evaluated in terms of computation time, accuracy, and several statistical indicators. [ABSTRACT FROM AUTHOR]

The fintech segment is currently one of the most rapidly growing industries, attracting numerous investors who anticipate substantial returns in the future. Notably, not only individual retail investors but also mutual fund agencies are actively engaged in predicting stock prices within this sector to maximize their trading gains. The purpose of the study is to formulate stock forecasting models for top three Fintech Companies of India i.e., Policy Bazar, One 97 Communications Paytm Ltd., and Niyogin Ltd. Using Random Forest model with high-frequency data in Python. The literature review section also proves that this study is a novel piece of work as none of the existing research study focused on predicting stock prices of Fintech Companies of India using Random Forest model. The data is extracted from www.moneycontrol.com and www.kotaksecurities.com, for the period from 1st October, 2022–30 th September, 2023. The study deals about 293,280 data points i.e., 3 companies @ 97,760 each. It has been found that the forecasting model of random forest provides very successful results for prediction as the co-efficient of determination of all the selected companies is more than 95%. [ABSTRACT FROM AUTHOR]

This study sought to develop models for predicting near-term (1–3 day) fecal contamination events in coastal shellfish growing waters. Using Random Forest regression, we (1) developed fecal coliform (FC) concentration models for shellfish growing areas using watershed characteristics and antecedent hydrologic and meteorologic observations as predictors, (2) tested the change in model performance associated when forecasted, as opposed to measured, rainfall variables were used as predictors, and (3) evaluated model predictor importance in relation to shellfish sanitation management criteria. Models were trained to 10 years of coastal FC measurements (n = 1285) for 5 major shellfish management areas along the Florida (USA) coast. Model performance varied between the 5 management areas with R2 ranging from 0.36 to 0.72. Antecedent precipitation variables were among the most important predictors in the day-of forecast models in all management areas. When forecasted rainfall was included in the models, wind components became increasingly important. • 1–3 day fecal coliform forecast models were trained with regulatory shellfish data. • 5 variations of Random Forest models for 5 management areas were evaluated. • Predictive performance ranged from R2 of 0.36 to 0.72. • Using forecasted rainfall as a predictor did not improve model performance. • Variable importance differed between models, implying varied prediction mechanisms. [ABSTRACT FROM AUTHOR]

Background: Depression and post-traumatic stress disorder (PTSD) are among the most common psychiatric disorders observed in children and adolescents exposed to sexual abuse.Objective: The present study aimed to investigate the effects of many factors such as the characteristics of a child, abuse, and the abuser, family type of the child, and the role of social support in the development of psychiatric disorders using machine learning techniques.Participants and Settings: The records of 482 children and adolescents who were determined to have been sexually abused were examined to predict the development of depression and PTSD.Methods: Each child was evaluated by a child and adolescent psychiatrist in the psychiatric aspect according to the DSM-V. Through the data of both groups, a predictive model was established based on a random forest classifier.Results: The mean values and standard deviation of the 10-k cross-validated results were obtained as accuracy: 0.82% (+/- 0.19%), F1: 0.81% (+/- 0.19%), precision: 0.81% (+/- 0.19%), recall: 0.80% (+/- 0.19%) for children with depression; and accuracy: 0.72% (+/- 0.12%), F1: 0.71% (+/- 0.12%), precision: 0.72% (+/- 0.12%), recall: 0.71% (+/- 0.12%) for children with PTSD, respectively. ROC curves were drawn for both, and the AUC results were obtained as 0.88 for major depressive disorder and 0.76 for PTSD.Conclusions: Machine learning techniques are powerful methods that can be used to predict disorders that may develop after sexual abuse. The results should be supported by studies with larger samples, which are repeated and applied to other risk groups. [ABSTRACT FROM AUTHOR]

The stopping power of a material upon interaction with an energetic ion is the key measure of how far that ion will travel. The implications of accurate particle range calculations are tremendous, affecting every single application in which particle radiation is involved, from nuclear power to medicine. An approach is presented which attempts to overcome current shortcomings in the theoretical understanding of stopping power, as well as the methods used to interpret and exploit measured data. This is a considerable challenge, however the use of a novel machine learning methodology is shown to hold great promise in this endeavour: the ultimate aim being the ability to correctly predict the stopping value for any energy, ion and target combination, having no pre-existing experimental data. A random forest regression algorithm is trained using over 34,000 experimental measurements, representing stopping power values for 522 ion-target combinations across the energy range 10 - 3 to 10 2 MeV/amu, and ion and target atomic masses 1 to >240. Evaluation is carried out using several fundamental error metrics, over the whole dataset as well as for individual combinations, to provide the most comprehensive understanding of performance when tested under strict cross-validation criteria. The resulting model is shown to yield predicted stopping power curves corresponding closely to those of the true experimental values, with an ability to generalise across target elements, compounds, mixtures, alloys and polymers, irrespective of phase, and for a wide range of ion masses. [ABSTRACT FROM AUTHOR]

The sneaker resale industry has reached the value of $2 billion and it is expected to increase by 200% in the next 5 years. This has changed the purpose of sneakers from a wardrobe collection into a promising business opportunity. The aim of this study is to compare several algorithms and decide which one has a better performance in predicting sneaker resale prices. Different techniques like linear regression and random forest have been used to make the predictions using sneaker sales history data gathered from StockX. It turns out that both models produce an outstanding result with similar values. However, with further evaluation, it can be concluded that random forest has a better performance compared to linear regression in predicting sneaker resale price. [ABSTRACT FROM AUTHOR]

The transmission dynamics of COVID-19 depend largely on the structure of the underlying contact network and individuals susceptibility. The latter factor is a primary concern of Covid-19 since recent studies reported conflicting conclusions regarding the characteristics of susceptible individuals. A susceptibility classification model of individuals is of great importance to governments and decision makers, as it facilitates imposing personalized protective strategies, which can save lives and minimize social and economic consequences. Machine learning models have been successfully used to predict individual susceptibility to different diseases such as cancer and visceral fat associated diseases. Existing contributions that predict susceptibility of COVID-19 are limited and use a large number of features including clinical data and medical history, which may be hard to obtain or unavailable. In this work, we investigate the use of machine learning models, namely multilayer perceptron (MLP), support vector machine (SVM), decision tree (DT), and random forest (RF) to predict susceptibility of individuals based on demographic data, including age, gender, nationality, and location. We use a dataset of test records of all individuals who were admitted to take nasopharyngeal swabs for COVID-19 in Saudi Arabia between March 2, 2020 until April 25, 2020. Our experiments show promising results with prediction accuracies of 85.6%, 85.3%, 77.2%, and 74.2% using DT, RF, MLP, and SVM, respectively. [ABSTRACT FROM AUTHOR]

The scale-up of nanocellulose production, a still challenging task, could benefit greatly from machine learning. In this paper, random forest regressors were applied to predict two important parameters for nanocellulose processing: nanofibrillation yield and aspect ratio. The dataset comprised diverse pulp sources, ranging from non-woody and woody feedstock, and two different types of pulp pre-treatments, namely mechanical refining and enzymatic hydrolysis. The obtained models were evaluated in terms of number of features and optimized with regard to some important hyperparameters, using the Bayesian optimization function from the SkOpt library. Considering the heterogeneity of the data, the prediction of both aspect ratio and yield of nanofibrillation was deemed satisfactory. For the former, R² scores of 0.986, 0.902 and 0.932 were obtained for the training, validation and testing subsets, respectively. Regarding the latter, R² scores of 0.994, 0.947 and 0.877 were achieved for the same types of subsets. Other important remarks are that enzyme dosage was the most important feature for predicting aspect ratio, while it did not play an important role for yield modelling. For this output, the transmittance at 600 nm accounted for almost 90% of the total feature importance. [Display omitted] • Aspect ratio and nanofibrillation yield were predicted by random forest regressors. • A 140-point dataset was gathered from experiments from different woods and crops. • Both independent variables and directly measurable responses constituted the inputs. • Bayesian optimization helped attain R2 of 0.90–0.95 for the validation subsets. • The kind of feedstock had lower feature importance than the type of pretreatment. [ABSTRACT FROM AUTHOR]

The integration of solar energy into power systems is essential for the future sustainability of power systems, particularly for isolated systems, such as microgrids, where establishing a primary transmission network is difficult. Therefore, the development of prediction methods becomes crucial to enable accurate forecasting of solar energy generation, facilitating efficient planning and operation of these systems and ensuring their long-term viability. This study proposes distinct forecasting models for solar irradiance forecasting: an autoregressive (AR) model, a Random Forest model, and a Long Short-Term Memory (LSTM) neural network. The methodology involves preprocessing the historical solar irradiance data and performing feature engineering to extract relevant input features. The architectural design, hyperparameter tuning, and training procedures of each model are discussed in detail. The findings indicate that the LSTM model exhibits enhanced performance compared to the AR model, while maintaining similar predictive accuracy to the Random Forest model in forecasting global solar irradiance. Both models yield a mean absolute percentage error of roughly 25%, with the LSTM exhibiting the lower error rate. Moreover, the LSTM model showcases an advancement over the AR model, resulting in a reduction of approximately 10 W/m 2 for both root mean square error and mean absolute error. This finding highlights the effectiveness of LSTM networks in capturing long-term dependencies for accurate solar irradiance forecasting. Furthermore, an analysis of the models' interpretability is conducted, offering valuable insights into the key factors that contribute to the shaping of solar irradiance patterns. These insights hold practical significance for the optimization of renewable energy systems. • Solar irradiance prediction compared in Ecuador cities: AR model, Random Forest, LSTM. • LSTM model improves over AR, comparable to Random Forest. • LSTM networks capture long-term dependencies for accurate solar forecasting. • Model interpretability provides insights for optimizing renewable energy systems. [ABSTRACT FROM AUTHOR]

Abstract Three methods, smart persistence, artificial neural network and random forest, are compared to forecast the three components of solar irradiation (global horizontal, beam normal and diffuse horizontal) measured on the site of Odeillo, France, characterized by a high meteorological variability. The objective is to predict hourly solar irradiations for time horizons from h+1 to h+6. The random forest (RF) method is the most efficient and forecasts the three components with a nRMSE from 19.65% for h+1–27.78% for h+6 for the global horizontal irradiation (GHI), a nRMSE from 34.11% for h+1–49.08% for h+6 for the beam normal irradiation (BNI); a nRMSE from 35.08% for h+1–49.14% for h+6 for diffuse horizontal irradiation (DHI). The improvement brought by the use of RF compared to Artificial Neural Network (ANN) and smart persistence (SP) increases with the forecasting horizon. A seasonal study is realized and shows that the forecasting of solar irradiation during spring and autumn is less reliable than during winter and summer because during these periods the meteorological variability is more important. Highlights • Forecasting of the three solar components: beam, diffuse and global. • Comparison of three forecasting methods: Smart persistence, MLP and Random Forest. • Good accuracy of the random forest forecasting tool. • Forecasting of hourly solar data with a time horizon from h+1 to h+6. [ABSTRACT FROM AUTHOR]

When a forecast of the total value over several time periods ahead is required, forecasters are presented with two temporal aggregation (TA) approaches to produce required forecasts: i) aggregated forecast (AF) or ii) aggregate data using non-overlapping temporal aggregation (AD). Often, the recommendation is to aggregate data to a frequency relevant to the decision the eventual forecast will support and then produce the forecast. However, this might not be always the best choice and we argue that both AF and AD approaches may outperform each other in different situations. Moreover, there is a lack of evidence on what indicators may determine the superiority of each approach. We design and execute an empirical experiment framework to first explore the performance of these approaches using monthly time series of M4 competition dataset. We further turn the problem into a classification supervised learning by constructing a database consisting of features of each time series as predictor and model class labelled as AF/AD as response/outcome. We then build machine learning algorithms to investigate the association between time series features and the performance of AF and AD. Our findings suggest that both AF and AD approaches may not consistently generate accurate results for every individual series. AF is shown to be significantly better than AD for the monthly M4 time series, especially for longer horizons. We build several machine learning approaches using a set of extracted time series features as input to predict accurately whether AD or AF should be used. We find out that Random Forest (RF) is the most accurate approach in correctly classifying the outcome assessed both by statistical measures such as misclassification error, F-statistics, area under the curve, and a utility measure. The RF approach reveals that curvature, nonlinearity, seas_pacf, unitroot_pp, mean, ARCHM.LM, Coefficient of Variation, stability, linearity, and max_level_shif are among the most important features in driving the predictions of the model. Our findings indicate that the strength of trend, ARCH.LM, hurst, autocorrelation lag 1, unitroot_pp, and seas_pacf may favour AF approach, while lumpiness, entropy, nonlinearity, curvature, and strength of seasonality may increase the chance of AD performing better. We conclude the study by summarising the findings and present an agenda for further research. [ABSTRACT FROM AUTHOR]

Predicting future land change is crucial in anticipating societal and environmental impacts and informing responses at different scales. We designed an integrated, high-resolution, land-change model and forecasted Australia's land change for the years 2020, 2025 and 2030 for Cropland, Forest, Grassland, and Built-up land-uses using cloud-based and high-performance computing. A spatially explicit set of drivers was fed into a random forest classifier to generate 30-m per-class suitability layers for the country, which were then used for allocating land-use. The model was validated against 2015 data, then land-use was projected until 2030. Accuracy at the national level was ∼94%. Forecasts showed increases in Grassland and Built-up areas and decreases in Forest and Cropland. Our modelling framework expands the current capabilities of large-scale land-change models and provides a first-of-its-kind multiclass land forecast for Australia that can inform land policy at multiple scales in Australia. • We developed a continental-scale multiclass land-change model at 30-m resolution. • Land change for Australia was forecast for 2020, 2025 and 2030. • Overall accuracy at national level was 93.8% and not lower than 86% at State level. • Grassland and Built-up areas increased, while Forest and Cropland areas decreased. • The outputs provide new insights into future Australian land trajectories. [ABSTRACT FROM AUTHOR]

Highlights • Machine learning models are compared for simulation and forecasting of streamflows. • Support Vector Regression, Artificial Neural Networks and Random Forest are used. • A base flow separation method is coupled with machine learning models. • North Fork, Chehalis, Carson and Sacramento rivers in the United States are analysed. • Base flow separation method improved streamflow simulation to a certain degree. Abstract Efficient simulation of rainfall-runoff relationships is one of the most complex problems owing to the high number of interrelated hydrological processes. It is well-known that machine learning models could fail in simulating streamflows from only meteorological variables in the absence of antecedent streamflow values. The main reason for this could be low and lagged relationships between streamflow and meteorological variables. To overcome this inefficiency, for the first time, we developed a simulation framework by coupling a base flow separation method to three machine learning methods. It was demonstrated that separating streamflow into different components such as base flow and surface flow can be useful for improving simulation and forecasting capabilities of machine learning models. We simulated streamflow in four rivers in the United States with Support Vector Regression (SVR), Artificial Neural Networks (ANNs) and Random Forest (RF) as a function of precipitation (P), temperature (T) and potential evapotranspiration (PET). We concluded that the base flow separation method improved the simulation performances of the machine learning models to a certain degree. Apart from the simulation scheme, we also employed a forecasting scheme by using the antecedent observed discharge values in addition to P, T, and PET. We discussed performances of models in simulation and forecasting of streamflow regarding model types, input structures and catchment dynamics in detail. [ABSTRACT FROM AUTHOR]

Fuel consumption influences both the economic and environmental perspectives of shipping. With the help of machine learning, meaningful knowledge and complex relationships can be extracted from high-dimensional historical data. In this study, machine learning models were developed to predict the fuel consumption of harbour vessels with ship-related and meteorological factors. The superiority of machine learning models over statistical linear regression model (Ridge regression) has been proved. This study further investigated whether the use of meteorological factors enhances the prediction of fuel consumption. A case study on the prediction of tugboat fuel consumption was conducted. The Random Forest model outperformed the other models. Comparative experiments showed that meteorological factors collectively add value to the fuel consumption prediction, which enhances the accuracy from 0.7 to 38.9%. The potential uses of the prediction results are highlighted in terms of both operational management and environmental evaluation aspects. • Machine learning models were developed to predict the fuel consumption of harbour vessels. • The superiority of machine learning models over statistical linear regression model (Ridge regression) has been proved. • The Random Forest model outperformed the other models. • The meteorological factors collectively add value to fuel consumption prediction, which enhances the accuracy from 0.7 to 38.9%. [ABSTRACT FROM AUTHOR]

• Precipitation δ 18O predicted using geostatistical and machine learning algorithms. • Prediction performance evaluated using ordinary and hybrid error metrics. • Machine learning algorithms performed better, particularly using scarcer datasets. • Two spatiotemporally different subcontinental datasets of precipitation δ 18O were used. Stable isotopes of precipitation are important natural tracers in hydrology, ecology, and forensics. The spatially explicit predictions of oxygen and hydrogen isotopes in precipitation are obtained through different interpolation techniques. In the present study we aim to examine the performance of various interpolation techniques when predicting the spatial distribution of precipitation stable isotopes. The efficiency of combined geostatistical tools (i.e. regression kriging; RK) and various machine learning methods (including regression enhanced random forest methods: MRRF, RERF) are compared in interpolating the spatial variability of precipitation stable oxygen isotope values from two different sampling networks in Europe. To assess the performance of the models, mean squared error (MSE), nonparametric Kling Gupta efficiency (KGE), absolute differences and relative mean absolute error metrics were employed. It was found that the combination of the different regression techniques with Random Forest can produce estimations with comparable accuracy in terms of descending order of overall average MSE, MRRF: 2.61, RK: 2.77, RERF: 2.99, RF: 3.08. The best performing combined random forest model variant (MRRF) outperformed regression kriging in terms of a hybrid error metric (KGE) by 7.5%. Sequential random rarefying the station networks showed that machine-learning methods are more capable of maintaining high prediction accuracy even with fewer input data. This can be a great advantage when a suitable method is needed to predict the stable isotope composition of precipitation for large spatial domains where the spatial density of data stations shows large differences. [ABSTRACT FROM AUTHOR]

Due to inherent errors in the chemical transport models, inaccuracies in the input data, and simplified chemical mechanisms, ozone (O 3) predictions are often biased from observations. Accurate O 3 predictions can better help assess its impacts on public health and facilitate the development of effective prevention and control measures. In this study, we used a random forest (RF) model to construct a bias-correction model to correct the bias in the predictions of hourly O 3 (O 3 -1h), daily maximum 8-h O 3 (O 3 -Max8h), and daily maximum 1-h O 3 (O 3 -Max1h) concentrations from the Community Multi-Scale Air Quality (CMAQ) model in the Yangtze River Delta region. The results show that the RF model successfully captures the nonlinear response relationship between O 3 and its influence factors, and has an outstanding performance in correcting the bias of O 3 predictions. The normalized mean biases (NMBs) of O 3 -1h, O 3 -Max8h, and O 3 -Max1h decrease from 15.8%, 20.0%, and 17.0.% to 0.5%, −0.8%, and 0.1%, respectively; correlation coefficients increase from 0.78, 0.90, and 0.89 to 0.94, 0.95, and 0.94, respectively. For O 3 -1h and O 3 -Max8h, the original CMAQ model shows an obvious bias in the central and southern Zhejiang region, while the RF model decreases the NMB values from 54% to −1% and 34% to −4%, respectively. The O 3 -1h bias is mainly caused by the bias of nitrogen dioxide (NO 2). Relative humidity and temperature are also important factors that lead to the bias of O 3. For high O 3 concentrations, the temperature bias and O 3 observations are the major reasons for the discrepancy between the model and the observations. [Display omitted] • The RF model greatly improves accuracy of hourly O 3 predictions from CMAQ. • The RF model significantly reduced O 3 bias in the central and southern Zhejiang. • O3-1h bias is mainly caused by the bias of nitrogen dioxide. • Relative humidity and temperature have important impacts on the O 3 bias. [ABSTRACT FROM AUTHOR]

[Display omitted] • Machine learning methods were employed to predict bed expansion ratio. • XGBoost model has good regression and prediction ability with R2 higher than 0.97. • The relative importance of operating variables on bed expansion ratio was obtained. • Particle size has the strongest effect on bed expansion ratio. Liquid–solid fluidizations are mainly affected by particle density, particle size, particle shape, superficial velocity, liquid viscosity, etc. Owing to these complex high-dimensional impact processes, rapid and accurate prediction of fluidization characteristics is necessary but, still challenging. In this study, machine learning models were developed and employed to predict bed expansion ratio. Machine learning models, namely, linear, random forest, and XGBoost regression models, were trained using a measured 183- bed expansion ratio dataset under different operating conditions. By comparison, the XGBoost model had strongest regression and prediction ability with R2 higher than 0.97 and the predictions were very time-saving with calculation time less than 1 s. Finally, the bed expansion characteristics were analyzed and the operating conditions were evaluated by the developed model. The relative importance of operating conditions on bed expansion ratios was particle size > particle density > superficial velocity. [ABSTRACT FROM AUTHOR]

This paper analyzes the performance of three types of statistical models and a well-known physics-based model for forecasting the wave energy flux. The forecasts are run over horizons of 1–24 h at five buoys located in the Bay of Biscay. The data resolution is hourly. The 1999–2005 timeframe is used to train the models. The forecasts are run and evaluated over the six-year period from 2006 to 2012. The statistical forecasting models use three techniques: analogues, random forests (a machine learning algorithm) and a combination of the two. The physics model is the Wave Model (WAM). The forecasts are compared at a 95% confidence level with the simplest prediction – Persistence – and also with the nearest grid point of the WAM forecasts. Over horizons between 3 and 16–19 h at locations near the coast (where wave farms may be installed), the random forests models outperform the others, including WAM and Persistence. These models exploit the inherent predictability associated with the strong autocorrelation present in ocean energy values. The additional prognostic capabilities that random forests models provide over Persistence, are due to their ability to sucessfully incorporate the information that both, atmospheric and sea-state variables provide. [ABSTRACT FROM AUTHOR]

Non-traditional safety (NTS) events pose an increasingly serious threat to the security of sea lanes, highlighting the significance of prediction and prevention. In this paper, a random forest (RF)-based approach is provided so as to assess the severity and risk of NTS events in sea lanes. Data was manually collected from the Global Integrated Shipping Information System (GISIS) and first classified into four severity levels, according to the classification criteria from the International Maritime Organization (IMO) Casualties and Incidents reports. The risk factors are determined based on extensive literature review and data analysis. Then, a random oversampling (RO) method is used to process the unbalanced samples before using them as an input in the RF, where they are divided into a training set for rule learning and a test set for verifying the model's validity. Finally, the risk is calculated based on the severity probabilities returned by the RF-RO, and the severity is quantified by expert scoring. The results reveal that the proposed RF-RO performs with the greatest accuracy among similar algorithms. The feature importance measurement indicates that time and ship type are the most influential risk factors in anticipating an NTS. These findings provide a reference for stakeholders in implementing security management. [ABSTRACT FROM AUTHOR]

Prediction of indoor airborne pollutant concentrations can enable a smart indoor air quality control strategy that potentially reduces building energy use and improves occupant comfort. In service of this overarching goal, this work pursues four objectives: 1) Determine which low-cost airborne pollutant sensors are useful for prediction of indoor air quality variables of interest, investigating whether a few commercially available sensors held value for making such predictions. 2) Investigate which algorithms are most useful for making these predictions. 3) Develop an understanding of how far into the future we can conceivably predict indoor concentrations based on low-cost airborne pollutant signals. 4) Investigate methods for predicting elevated concentration events from historical data. Four different methods (Rolling Average, Random Forest, Gradient Boosting, and Long-Short Term Memory) for predicting eight indoor pollutant concentrations (carbon dioxide, nitrogen dioxide, ozone, PM 1, PM 2.5, PM 10, formaldehyde, total volatile organic compounds) are compared for their ability to predict future sensor signals in a single commercial building in California. Long-Short Term Memory was consistently the best method for predicting indoor pollutants, though the best combinations of input variables differed depending on pollutant of interest. To predict elevated concentration events, results show that indirect classification through a regression prediction that was then compared to a threshold performed marginally better than a direct classification prediction for all pollutants except PM 1. • Time series of indoor airborne pollutants predicted with machine learning • Most accurate predictive model is Long-Short Term Memory algorithm • Using Long-Short Term Memory algorithm around 90% of variance in data predicted by model • 50–80% of high concentration events predicted correctly for most pollutants • Accuracy of prediction falls and overfitting increases for greater than 1-h ahead forecasts [ABSTRACT FROM AUTHOR]

Short-term vessel speed prediction is exceedingly crucial to various issues in maritime transport, notably in cases with high precision requirement, such as trajectory tracking and collision avoidance. The methods for speed prediction have successively experienced the evolution of empirical formulas, physics-based approaches and data-driven models. Compared to the other two pioneering methods, building the model from data-driven learning seems to be a more appropriate way for ever-larger vessels that entered service in recent years. However, owing to insufficient speed-related features that can be known in advance, such as the engine state and propeller propulsion, traditional data-driven models usually have poor generalization ability when facing new test data. In this paper, taking a case study of a 400,000 DWT ore carrier sailing between China and Brazil, we first provide comparisons of several widely-used data-driven models and select the framework of random forest as the basic. Then, by considering the hardship of directly utilizing real-time monitoring data, we creatively put forward a novel prediction model by introducing sequential data of last historical speed records to bring the potential information of engine state into model learning. Through comprehensive experiments, the superiority and robustness of the proposed model are demonstrated by comparing with other frequently-used time-series prediction methods. • Taking a case of very large ore carrier, data-driven models are analyzed in details. • We propose a new model by introducing the time-series data into the random forest. • The effectiveness and superiority of the proposed model has been demonstrated. [ABSTRACT FROM AUTHOR]

• We apply machine learning to forecast the Olympic medal distribution. • We are first to consistently outperform the more traditional naïve forecast. • We also improve the forecasting accuracy presented in more recent work. Forecasting the number of Olympic medals for each nation is highly relevant for different stakeholders: Ex ante, sports betting companies can determine the odds while sponsors and media companies can allocate their resources to promising teams. Ex post, sports politicians and managers can benchmark the performance of their teams and evaluate the drivers of success. We apply machine learning, more specifically a two-staged Random Forest, to a dataset containing socioeconomic variables of 206 countries (1991–2020). For the first time, we outperform the more traditional naïve forecast for four consecutive Olympics between 2008 and 2020. [ABSTRACT FROM AUTHOR]

[Display omitted] • We built a random forest model on ozone using TROPOMI and HRRR data. • Our model had a cross-validated (CV) R 2 of 0.84 for daily ground ozone concentrations. • Model predictions captured daily ozone distributions at 10 km spatial resolution. • TROPOMI column O 3 data improved the characterization of high concentrations. Estimating ground-level ozone concentrations is crucial to study the adverse health effects of ozone exposure and better understand the impacts of ground-level ozone on biodiversity and vegetation. However, few studies have attempted to use satellite retrieved ozone as an indicator given their low sensitivity in the boundary layer. Using the Troposphere Monitoring Instrument (TROPOMI)'s total ozone column together with the ozone profile information retrieved by the Ozone Monitoring Instrument (OMI), as TROPOMI ozone profile product has not been released, we developed a machine learning model to estimate daily maximum 8-hour average ground-level ozone concentration at 10 km spatial resolution in California. In addition to satellite parameters, we included meteorological fields from the High-Resolution Rapid Refresh (HRRR) system at 3 km resolution and land-use information as predictors. Our model achieved an overall 10-fold cross-validation (CV) R 2 of 0.84 with root mean square error (RMSE) of 0.0059 ppm, indicating a good agreement between model predictions and observations. Model predictions showed that the suburb of Los Angeles Metropolitan area had the highest ozone levels, while the Bay Area and the Pacific coast had the lowest. High ozone levels are also seen in Southern California and along the east side of the Central Valley. TROPOMI data improved the estimate of extreme values when compared to a similar model without it. Our study demonstrates the feasibility and value of using TROPOMI data in the spatiotemporal characterization of ground-level ozone concentration. [ABSTRACT FROM AUTHOR]

• Systematic framework is established that formalizes the scope of the smart grids. • A deep learning-based approach is proposed for predicting load in smart grids quite accurately with use of features of various neural networks for efficient load balancing in smart grids. • Use of the proposed approach to efficiently determine the load forecasting characteristics in macro grids and microgrids. • Comparative analysis of proposed work with state-of-the-art for load forecasting and feature extraction in smart grids. Load forecasting plays an essential role in effective energy planning and distribution in a smart grid. However, due to the unpredictable and non-linear structure of smart grids and large datasets' complex nature, accurate load forecasting is still challenging. Statistical techniques are being used for a long time for load forecasting, but it is inefficient. This paper tries to resolve challenges imposed by conventional methods like mean and mode by suggesting an ANN model for accurate load forecasting. Specifically, the LSTM and random forest approach has been used here. We compared our model to other models that use similar parameters and found that ours is more reliable and can be used for long-term forecasting. Our model has achieved an average overall accuracy of 96% and an average MSE of 4.486 with average CPU time consumption of 904.47 s, 872.43 s, and 908.32 s, respectively. Hence, the present model outperforms other existing methods. [ABSTRACT FROM AUTHOR]

Urban energy consumption is expected to continuously increase alongside rapid urbanization. The building sector represents a key area for curbing the consumption trend and reducing energy-related emissions by adopting energy efficiency strategies. Building age acts as a proxy for building insulation properties and is an important parameter for energy models that facilitate decision making. The present study explores the potential of predicting residential building age at a large geographical scale from open spatial data sources in eight municipalities in the German federal state of North-Rhine Westphalia. The proposed framework combines building attributes with street and block metrics as classification features in a Random Forest model. Results show that the addition of urban fabric metrics improves the accuracy of building age prediction in specific training scenarios. Furthermore, the findings highlight the way in which the spatial disposition of training and test samples influences classification accuracy. Additionally, the paper investigates the impact of age misclassification on residential building heat demand estimation. The age classification model leads to reasonable errors in energy estimates, in various scenarios of training, which suggests that the proposed method is a promising addition to the urban energy modelling toolkit. • Automatic prediction of age class for residential buildings. • Assessment of the importance of street and urban block attributes for building age prediction. • Investigation of the spatial autocorrelation patterns of building attributes. • Analysis of the transferability of model learning between geographical regions. • Quantifying the impact of age classification errors on building heat demand estimation. [ABSTRACT FROM AUTHOR]

Trait-based approaches offer complementary views to the classic taxonomic approach, which is a crucial step forward to unveil mechanisms of community assembly, species interactions, ecosystem functioning, and tackling important conservation issues. These approaches require an enormous sampling effort to provide complete trait datasets, consequently, missing data are very common. We evaluated the performance of the missForest algorithm, that uses the Random Forest method to impute species traits values using phylogenetic information. We simulated datasets with different sizes and proportions of missing data, different levels of trait conservatism, and trait correlation. We tested trait imputation using the missForest algorithm without phylogenetic information and adding the phylogenetic relationship among species, using phylogenetic eigenvector. Our results show that the level of phylogenetic signal in traits and the correlation among them are the main parameters that influence the measures of imputation error. The measures of errors are smaller when traits have higher levels of correlation and when traits are conserved in the phylogenetic tree. In general, the inclusion of phylogenetic vectors as predictors in the missForest algorithm improves the estimation of missing values. However, the importance of phylogenetic information to the imputation process depends on the proportion of missing entries, phylogenetic conservatism of traits, and the correlation among traits. The missForest algorithm seems to be a robust method for trait imputation, and it can be used to estimate missing traits without the exclusion of species. Thus, we expect to have contributed with a new step to guide methodological choices to impute entire databases of traits with the goal of decreasing uncertainties and bias in the interpretation of ecological patterns and processes at different levels of ecological organization. • Trait-based approaches offer complementary views to classic ecological studies. • Missing data are fairly common and not the exception in trait datasets. • The performance of Random Forest in imputing species trait values was evaluated. • The inclusion of phylogenetic information improves the estimation of missing values. • Trait correlation structure determine the accuracy of missing value prediction. [ABSTRACT FROM AUTHOR]

The detonation heat is the main performance index of ammunition used in air-to-air and underwater weapons. In the present work, a method for predicting the detonation heat performance of nitrogen-containing compounds based on RF combined QSAR is established. Firstly, the molecular descriptors of nitrogen-containing compounds and the corresponding detonation heat performance parameters are calculated by Gaussian 09 and Codessa 3.2 software, and normalization was used for eliminating orders of difference between the unit and order of magnitude. On the basis, the variable importance (VI) was applied to select the feature variables in the molecular descriptor, and the variable importance threshold was optimized based on out of bag (OOB) error estimation. Then, the optimized feature variables were compression by wavelet transform (WT), and the obtained result was compared with the RF models based on VI. Under the optimized characteristic molecular descriptors, an RF-QSAR relationship model based on the prediction of the detonation heat performance of nitrogen-containing compounds was constructed, and the detonation heat performance of nitrogen-containing compounds in the prediction set was predicted. The results show that the wavelet transform-variable importance-random forest-quantitative structure-activity relationship (WT-VI-RF-QSAR) model can obtain better prediction performance and can more accurately predict the detonation heat performance of nitrogen-containing compounds. The correlation coefficient of calibration set (R2 C) is 0.9652, the mean relative error (MREC) is 9.03%, and the R2 P of prediction set is 0.8801, MREP is 10.52%. Compared with traditional experimental testing methods, this method has the advantages of fast analysis speed, accurate prediction results, and short monitoring period, etc. This research provides new ideas and methods for the performance prediction of energetic compounds and the design of new energetic materials. The prediction of the detonation heat of nitrogen compounds via random forest (RF) and a hybrid variable selection method. [Display omitted] • A method for predicting the detonation heat performance of nitrogen-containing compounds based on RF-QSAR is established. • A mixed variable selection method based on WT-VI was used to improve the predictive performance of RF model. • Compared with traditional methods, the method has the merits of fast analysis speed, precise prediction result, and safe. [ABSTRACT FROM AUTHOR]

[Display omitted] • Random Forest and Deep Auto-Encoder are combined with Support Vector Regression. • RF-SVR and DAE-SVR indicate suitable performance for investigating the uncertainty of streamflow. • DDMs show better performance in simulation of monthly streamflow during wet seasons. Streamflow forecasting is an important component of water resources planning and management. Data-Driven Models (DDMs) are common approaches for streamflow prediction. DDMs try to obtain a mathematical relation between variables without any prior knowledge or assumption about the physical relation between them. Hence, these models are effective for complex process modeling. In this study, we use the Support Vector Regression (SVR) and Multiple Linear Regression (MLR) models for long-term streamflow prediction. Moreover, three approaches, Random Forest (RF), Deep Auto-Encoder (DAE), and Principle Component Analysis method (PCA) are employed as pre-processing methods and they are combined to SVR and MLR (six hybrid models produced) in order to improve the accuracy of prediction and reduce its uncertainty. Different evaluation criteria, including R2, RMSE, NSE, P-factor, and R-factor are computed to evaluate the efficiency of the hybrid models. The results of our study on Bookan dam indicate that the DAE-SVR model, in comparison with the PCA-MLR, RF-MLR, DAE-MLR, PCA-SVR hybrid models, provides a significant improvement in the runoff prediction accuracy and the reduction of its uncertainty. DAE-SVR and RF-SVR have the highest values in NSE (0.93 and 0.87, respectively), and the lowest values in RMSE (33.85 and 43.68 MCM, respectively) for the first forecast horizon. The uncertainty analysis of the streamflow prediction shows that DAE-SVR has the best performance at the 95% confidence level due to its P-factor and R-factor values (0.9 and 0.64, respectively). Furthermore, the DAE-SVR performance is likewise better than other hybrid models in the two, three, and four months forecast horizons. [ABSTRACT FROM AUTHOR]

• Anthocyanin derivatives' evolution pattern during wine aging was elucidated. • Color contribution of anthocyanin derivatives in various aging stage was clarified. • Pinotin, A-v-F and vitisin A contribute to wine prolonged aging color greatly. • Machine learning methods showed good predictive performance on wine aging stage. Anthocyanin derivatives and chromatic characteristics of 234 different-vintage red wine were investigated based on a targeted HPLC-MS/MS and CIELAB approach. The K-means cluster analysis showed that the evolution pattern varies amongst anthocyanin derivative classes. Their stabilities are: pinotins > flavanyl-pyranoanthocyanins, vitisin A > monomeric anthocyanin, direct anthocyanin-flavan-3-ols condensation products > vitisin B, anthocyanin ethyl-linked flavan-3-ols products. The proportion of most pyranoanthocyanins becomes more significant among all detected anthocyanin derivatives during wine aging, whereas flavanols-related anthocyanin derivatives (except for flavanyl-pyranoanthocyanins) decreased drastically. PLSR showed that aging tawny characteristics is related to pyranoanthocyanins except for vitisin B, especially pinotins, whereas monomeric anthocyanins and flavanol-related derivates (except for flavanyl-pyranoanthocyanins) contribute to red violet color. But aging color density is more associated with the content of vitisin A and flavanyl-pyranoanthocyanins. Two predictive models based on random forest and support vector machine modeling showed good performance in predicting the extent of wine aging. [ABSTRACT FROM AUTHOR]

People typically spend most of their time indoors. It is of importance to establish prediction models to estimate PM 2.5 concentration in indoor environments (e.g., residential households) to allow accurate assessments of exposure in epidemiological studies. This study aimed to develop models to predict PM 2.5 concentration in residential households. PM 2.5 concentration and related parameters (e.g., basic information about the households and ventilation settings) were collected in 116 households during the winter and summer seasons in Hong Kong. Outdoor PM 2.5 concentration at households was estimated using a land-use regression model. The random forest machine learning algorithm was then applied to develop indoor PM 2.5 prediction models. The results show that the random forest model achieved a promising predictive accuracy, with R2 and cross-validation R2 values of 0.93 and 0.65, respectively. Outdoor PM 2.5 concentration was the most important predictor variable, followed in descending order by the household marked number, outdoor temperature, outdoor relative humidity, average household area and air conditioning. The external validation result using an independent dataset confirmed the potential application of the random forest model, with an R2 value of 0.47. Overall, this study shows the value of a combined land-use regression and machine learning approach in establishing indoor PM 2.5 prediction models that provide a relatively accurate assessment of exposure for use in epidemiological studies. Image 1 • A practical indoor PM 2.5 prediction framework was proposed. • The combined land-use regression and machine learning approach was applied. • External validation was proposed to evaluate the prediction model. • The random forest model outperformed the linear mixed-effect regression model. • The outdoor PM 2.5 concentration is the most important predictor variable. [ABSTRACT FROM AUTHOR]

• Random forest was used for predicting sediment transport in sewer pipes. • The Random forest results outperformed the predictions of the previously reported models. • Overfitting was avoided with the new Random forest based model. • Volumetric sediment concentration is the most important parameter for predicting self-cleansing in sewers. Sediment transport in sewers has been extensively studied in the past. This paper aims to propose a new method for predicting the self-cleansing velocity required to avoid permanent deposition of material in sewer pipes. The new Random Forest (RF) based model was implemented using experimental data collected from the literature. The accuracy of the developed model was evaluated and compared with ten promising literature models using multiple observed datasets. The results obtained demonstrate that the RF model is able to make predictions with high accuracy for the whole dataset used. These predictions clearly outperform predictions made by other models, especially for the case of non-deposition with deposited bed criterion that is used for designing large sewer pipes. The volumetric sediment concentration was identified as the most important parameter for predicting self-cleansing velocity. Image, graphical abstract [ABSTRACT FROM AUTHOR]

Predicting long-term spatiotemporal characteristics of fine particulate matter (PM 2.5) is important in China to understand historical levels of PM 2.5 , to support health effects research of both long-term and short-term exposures to PM 2.5 , and to evaluate the efficacy of air pollution control policies. Satellite-retrieved aerosol optical depth (AOD) provides a unique opportunity to characterize the long-term trends of ground-level PM 2.5 at high spatial resolution. However, the missing rate of AOD in Northeastern China (NEC) is very high, especially in winter, and challenges the accuracy of long-term predictions of PM 2.5 if left unresolved. Using random forest algorithms, this study developed a gap-filling approach combing satellite AOD, meteorological data, land use parameters, population and visibility in the NEC during 2005–2016. The model, including all predictors, combined with a model without AOD was able to fill the gap of PM 2.5 predictions caused by missing AOD at 1-km resolution. The R2 (RMSE) of the full-coverage predictions was 0.81 (18.5 μg/m3) at the daily level. Gap-filled PM 2.5 predictions on days with missing AOD reduced the relative prediction error from 28% to 2.5% in winter. The leave-one-year-out-cross-validation R2 (RMSE) of the full-coverage predictions was 0.65 (16.3 μg/m3) at the monthly level, indicating relatively high accuracy of predicted historical PM 2.5 concentrations. Our results suggested that AOD helped increase the reliability of historical PM 2.5 prediction when ground PM 2.5 measurements were unavailable, even though predictions from the AOD model only accounted for approximate 37% of the whole dataset. Predicted PM 2.5 level in NEC have increased since 2005, reached its peak during 2013–2015, then saw a major decline in 2016. Our high-resolution predictions also showed a south to north gradient and many pollution hot spots in the city clusters surrounding provincial capitals, as well as within large cities. Overall, by combining predictions from the AOD model with higher accuracy and predictions from the non-AOD model to achieve full coverage, our modeling approach could produce long-term, full-coverage historical PM 2.5 levels in high-latitude areas in China, despite the widespread and persistent AOD missingness. Unlabelled Image • AOD increased the accuracy of models for predicting historical PM 2.5 levels. • Predictions of PM 2.5 from gap-filling models achieved 100% coverage in NEC. • Full-coverage estimates corrected the prediction bias of PM 2.5 in winter of NEC. [ABSTRACT FROM AUTHOR]

• 1975 samples of varied soils were analyzed for fertility and elemental contents. • pXRF data was successfully used to rapidly predict soil fertility properties. • Random Forest outperformed other machine learning algorithms. • Prediction models allowed spatial distribution of soil fertility properties. • Inexpensive and environment-friendly soil fertility prediction was achieved by pXRF. Conventional soil fertility analyses are laborious, costly, time-consuming, and produce hazardous waste. The high demand of these laboratory-based analyses prompted us to investigate an environment-friendly, rapid, and inexpensive methodology for soil fertility assessment. Portable X-ray fluorescence (pXRF) spectrometry allows the determination of total elemental concentration in soils quickly, simply and without hazardous waste production. However, incipient usage of this technology for the prediction of soil fertility properties has been reported for tropical conditions. Soil samples were collected from seven Brazilian states (n = 1975) aiming to use pXRF data to predict contents of available or exchangeable Ca2+, Mg2+, Al3+, K+ and P by testing different algorithms using 70% of the samples for model training, and the remaining 30% for model validation. In addition to point data predictions, the best performing models were applied to data obtained from a farm within the studied regions with a known cropping history to create soil fertility maps and illustrate another applicability of this approach. The attested use of pXRF data and machine learning algorithms stepwise Generalized Linear Model (GLM) and Random Forest (RF) to predict the contents of relevant soil fertility properties exhibited great potential. Validation of the models corroborated that RF resulted in more accurate predictions than GLM. Validation R2 values ranged from 0.59 to 0.82. Maps created were coherent with expected distributions of soil fertility attributes. This environment-friendly methodology may be used for the assessment of soil fertility properties in a wide range of tropical and subtropical soils with minimum waste generation and reduced costs. [ABSTRACT FROM AUTHOR]

• SOC of 371 soil samples was predicted via Nix, PXRF-Rb, and texture. • Combining Nix color parameters and texture showed the best prediction of SOC. • Using PXRF-Rb as a proxy for clay could not produce satisfactory SOC prediction. • Color variables b* and a* extracted by Nix appeared influential in predicting SOC. The optimum ecosystem functioning is reliant on soil organic carbon (SOC) content which is traditionally measured in the laboratory via a cumbersome wet-chemistry method. This preliminary research tested whether a combination of Nix color sensor and portable X-ray fluorescence (PXRF) spectrometer data with soil texture data can improve soil SOC prediction accuracy relative to using them independently. A total of 371 samples representing diverse soil texture and SOC content were collected from three different ecoregions of eastern India: coastal saline zone, red and lateritic zone, and Gangetic alluvial zone. All dried, ground, and sieved samples were scanned via Nix and PXRF and random forest (RF) regression was used to predict soil SOC with different combinations of data. Soils were grouped into nine textural classes while soil SOC content exhibited substantial variability (0.08–2.26%). Comparing soil SOC with texture (sand + silt + clay), satisfactory prediction accuracy was observed (validation R2 = 0.70). Combining Nix extracted color parameters with texture substantially improved the model performance, producing the validation determination coefficient of 0.81. In contrast, PXRF-Rb, as a proxy of soil clay content was unable to achieve satisfactory prediction performance (R2 = 0.24), indicating the heterogeneity in soil mineralogical composition. The RF variable importance plot using Nix alone identified redness (a*) and yellowness to blueness (b*) as influential predictors, manifesting the impact of red color from Fe and Al-oxides and their significant negative correlation with soil SOC (r = −0.62 and −0.57 for a* and b*, respectively). These color parameters were again identified by the RF variable importance plot of (Texture + Nix)-model, implying that the SOC prediction improvement may be linked with the Nix sensor's capability of extracting useful information in the visible range. Summarily, a combination of Nix color variables and texture data was adept at predicting soil SOC in lieu of traditional laboratory analysis. The robustness of the (Texture + Nix)-based SOC prediction model can be augmented by incorporating more soil samples representing all 12 soil textural classes and variable SOC content, showing all possible soil colors in the pedogenic environment. [ABSTRACT FROM AUTHOR]

Approximately half of satellite aerosol retrievals are missing that limits the application of satellite data in PM 2.5 pollution monitoring. To obtain spatiotemporally continuous PM 2.5 distributions, various gap-filling methods have been developed, but have rarely been evaluated. Here, we reviewed and summarized four types of gap-filling strategies, and applied them to a random forest PM 2.5 prediction model that incorporated ground observations, chemical transport model (CTM) simulations, and satellite AOD for predicting daily PM 2.5 concentrations at a 1-km resolution in 2013 in the Beijing-Tianjin-Hebei region and the Yangtze River Delta. The model out-of-bag predictions were compared with national station measurements and external measurements to assess the performance of different gap-filling methods. We also conducted a by-city cross-validation and characterized the spatial distributions of PM 2.5 prediction when the AOD coverage was low. We found that the methods filling in missing data by regression, i.e. multiple imputation and decision tree, performed robustly to characterizing PM 2.5 variation at a high spatial resolution and the method filling in missing PM 2.5 predictions with decision tree overcame the problem of time-consuming computations. The method using spatiotemporal trends to fill in missing data, i.e. ordinary kriging and generalized additive mixed model, may be overrated in statistical evaluation tests, and predicted artificially oversmoothed PM 2.5 spatial distributions. We also revealed that CTM simulations benefited the prediction of PM 2.5 spatial distribution in all the models with various gap-filling strategies with higher prediction accuracy in the by-city cross-validation. We noticed that the PM 2.5 prediction was not sensitive to the resolution of CTM simulations and even the 12-km resolution CTM simulations benefited the high-resolution PM 2.5 prediction. · Four gap-filling strategies for high-resolution PM 2.5 predictions were evaluated. · Model performance was evaluated by both statistical tests and spatial distributions. · Filling missing PM 2.5 with decision tree balanced accuracy and computation time. · Inclusion of chemical transport model simulations benefited the PM 2.5 prediction. [ABSTRACT FROM AUTHOR]

• Intersections are among the most dangerous parts of the highway system. • Despite improved intersection control, intersection safety is still a major problem. • Real-time traffic and signal data were used to analyze vehicle occupants safety. • Study findings may assist in developing intersection safety countermeasures. Intersections are among the most dangerous roadway facilities due to the existence of complex movements of traffic. Most of the previous intersection safety studies are conducted based on static and highly aggregated data such as average daily traffic and crash frequency. The aggregated data may result in unreliable findings because they are based on averages and might not necessarily represent the actual conditions at the time of the crash. This study uses real-time event-based detection records, and crash data to develop predictive models for the vehicle occupants' injury severity. The three-year (2017–2019) data were acquired from the arterial highways in the City of Tallahassee, Florida. Random Forest (RF) and eXtreme Gradient Boosting (XGBoost) classifiers were used to identify the important factors on the vehicle occupants' injury severity prediction. The performance comparison of the two classifiers revealed that the XGBoost has a higher balanced accuracy score than RF. Using the XGBoost classifier, five topmost influential factors on injury prediction were identified. The factors are the manner of the collision, through and right-turn traffic volume, arrival on red for through and right-turn traffic, split failure for through traffic, and delays for through and right-turn traffic. Moreover, the partial dependency plots of the influential variables are presented to reveal their impact on vehicle occupant injury prediction. The knowledge gained from this study will be useful in developing effective proactive countermeasures to mitigate intersection-related crash injuries in real-time. [ABSTRACT FROM AUTHOR]

In China, the illegal adoption of missing persons and especially of missing children is a major public safety issue that affects social and family stability. Recent work has established a trafficking information network developed from a volunteer-managed database of missing persons that identifies and locates node cities and critical paths of illegal adoption. In order to evaluate locations where trafficking can be identified and provide direct advice for affected families, this study analyses the temporal and spatial distribution of the missing population and explores factors that affect their transfer. We use spatiotemporal information to construct multiple random forest (RF) models for predicting the locations of missing persons transfer on a larger spatial scale. The proposed independent RF models, namely, provinces potentially entered, destination grids, relative distances and relative directions models, achieve high levels of accuracy. Moreover, an integrated RF-based city-level prediction model can effectively locate the city a missing person was trafficked to. From our driving factor analysis, the transfer paths are strongly correlated with source provinces and grids. The study also shows that the transfer of missing persons is driven by multiple factors rather than by a single element. • The spatiotemporal distribution characteristics of the missing people were revealed. • A model is proposed to predict the current location of missing people. • The study can effectively help families determine the location of missing relatives. • This model can exploit the driving factors of missing population's transfer. [ABSTRACT FROM AUTHOR]

The straightforward prediction for the air-entry value of compacted soils is practically useful, but the investigation on this issue is scarce. This study presents three alternative straightforward prediction models for the air-entry value of compacted soils using the representative machine learning algorithms of multi expression programming (MEP), evolutionary polynomial regression (EPR) and random forest (RF). Five known soil properties (i.e. sand content, fines content, plasticity index, initial water content and initial void ratio) are used as input variables. All models are developed based on a large database, covering a wide range of soil classifications. The results show that all the three proposed models are appropriate to predict the air-entry values of different compacted soils, with high prediction accuracies for both the training and the testing data. The monotonicity, the sensitivity and the robustness of the three prediction models are evaluated, showing consistency among different models with a slight difference and providing a strong support for the model feasibility. On the whole, the MEP and the EPR models are recommended for more convenient applications with explicit expression, while higher prediction accuracy may require the RF model although no explicit expression can be derived. • Air-entry value of compacted soils is precited using three machine learning algorithms (MEP, EPR, RF) with high precision. • The correctness and the feasibility of all the three models are verified. • The MEP and the EPR models are recommended for more convenient applications with high prediction precision. [ABSTRACT FROM AUTHOR]

Evaluating the efficacy of air pollution control policies is an essential part of the decision-making process to develop new policies and improve existing measures. Since 2005, Fujian Province of Mainland China and Taiwan across the Taiwan Strait have both implemented aggressive air pollution control policies designed based on different principles, but a comprehensive evaluation of these control policies on PM 2.5 pollution levels is still lacking. In the current study, we assessed the effects of these policies in the Taiwan Strait Region from 2005 to 2018 using full-coverage, high-resolution PM 2.5 generated by a satellite-driven machine learning model. A ten-fold cross-validation for our prediction model showed an R 2 value of 0.89, demonstrating that these predictions can be used for policy evaluation. During the 14-year period, PM 2.5 levels in all areas of Fujian and Taiwan underwent a significant decrease. Separate regression models for policy evaluation in Taiwan and Fujian showed that all considered policies have mitigated PM 2.5 pollution to various degrees. The Clean Air Action Plans (CAAP) is the most effective control policy in Taiwan, while the Action Plan of Air Pollution Prevention and Control (APPC-AP) and Three-year Action Plan for Blue Skies (3YAP-BS) as well as their provincial implementation plans are the most successful in Fujian. The effectiveness of control policies, however, varies by land-use types especially for Taiwan. • PM 2.5 levels were estimated in the Taiwan Strait Region at 1-km resolution from 2005 to 2018 • The most effective control policies to lower PM 2.5 levels in Taiwan and Fujian were analyzed • The effectiveness of control policies varies by land use type in Taiwan [ABSTRACT FROM AUTHOR]

The present study aims to compare the performance of two machine learning techniques that can unveil the relationship between the input and target variables and predict the ground vibration (peak particle velocity, PPV) due to quarry blasting. To this end, a Random Forest (RF) model and a Bayesian Network (BN) model were developed. Before developing these models, and in order to illustrate the necessity of proposing new intelligent systems, a new empirical equation is proposed, using maximum charge per delay and distance from the blast-face. The results confirm that there is indeed a need to develop intelligent systems with more input parameters. Thus, a Feature Selection (FS) model is applied to decrease the dimensionality of data and remove the irrelevant data. The outputs of this technique set five parameters, hole depth, power factor, stemming, maximum charge per delay and distance from the blast-face, as the most important model inputs necessary to predict PPV. After constructing FS-BN and FS-RF models and comparing them under different conditions (i.e., computational cost, accuracy and robustness), it is found that the developed FS-RF model can be introduced as a new model in the field of blasting environmental issues. The accuracy level of the FS-RF model is quite high; 92.95% and 90.32% for the train and test stages, respectively, while 92.95% and 87.09% accuracy is calculated for train and test of the FS-BN model. Thus, both developed hybrid models can effectively unveil the relationships between the input and target variables. • We measured the most effective parameters on peak particle velocity induced by blasting. • A new peak particle velocity empirical equation was proposed and compared with previous equations. • A feature selection model was applied to remove the irrelevant predictors. • Two hybrid models were developed to predict peak particle velocity. • A comparison between these models was made and the best among them was chosen. [ABSTRACT FROM AUTHOR]

Different prediction models have important effects on the accuracy of spatial distribution simulations of heavy metals in soil. This study proposes a model (RFOK) combining a random forest (RF) with ordinary kriging (OK), multi-source environmental data such as terrain elements, site environmental elements, and remote sensing data were incorporated to predict the spatial distribution of heavy arsenic (As) in soil of a certain large arsenic slag site. The predictions results of RFOK were compared with those obtained using the RF, OK, inverse distance weighted (IDW), and stepwise regression (STEPREG) models for assessment of prediction accuracy. The results showed that arsenic pollution was widely distributed and the center of the site, including arsenic slag stacking area and production area were seriously polluted. The overall spatial distribution of arsenic pollution simulated by the five models was similar, but the IDW, RF, OK, and STEPREG showed less spatial variation of soil pollution, while RFOK simulation can better express the characteristics of details in change. The cross-validation results showed that RFOK had the lowest root-mean-square error (RMSE), mean absolute error (MAE), and mean relative error (MRE) relative to the other four models, followed by RF, OK, IDW, and STEPREG. The RMSE, MAE and MRE of RFOK decreased by 62.2%, 64.3% and 68.7%, respectively, relative to the RF model with the second highest accuracy. Compared with the traditional spatial distribution prediction model, the RFOK model proposed in this study has excellent spatial distribution prediction ability for soil heavy metal pollution with large spatial variation characteristics, which can fully explain the nonlinear relationship between pollutant content and its environmental impact elements. Prediction maps created using five models for As concentrations in soils at the contaminated site. Image 1 • The spatial distribution of soil As in a large arsenic slag site were investigated. • The RFOK model has excellent prediction ability for soil heavy metal pollution. • The nonlinear relationship between content and environmental elements was captured. • DisAs, Disroad and Disriver were the main factors influencing the arsenic in soil. [ABSTRACT FROM AUTHOR]