Random Regression Research Articles

Mapping the soil organic matter content in Northeast China considering the difference between dry lands and paddy fields

Mapping the soil organic matter (SOM) content of cultivated lands at the regional scale is of great significance for evaluating the cultivated land quality and monitoring the soil carbon cycle, especially in the fertile black-soil area of China. The large paddy fields area is one of the characteristics of the black-soil area in Northeast China. The vast differences between paddy fields and dry lands may pose a major challenge in mapping the SOM contents of local cultivated lands. In this study, the SOM of cultivated lands in Northeast China is taken as the research object, and all available Landsat-8 images from 2014 to 2022 and the main environmental covariates (climate and terrain) are obtained. By combining the random forest regression algorithm, SOM prediction models of paddy fields and dry lands are established to evaluate the optimal window period and appropriate environmental covariates for paddy fields and dry lands. Finally, the accuracy difference between the global regression and local regression results for distinguishing paddy fields and dry lands is compared. The results showed that (1) the SOM content in Northeast China increased gradually from south to north, and the average SOM content in paddy fields was approximately 0.4 % higher than that in dry lands; (2) the SOM mapping time windows in paddy fields and dry lands in Northeast China differed, with paddy fields mapped in April and dry lands mapped in May; (3) the addition of environmental covariates improved the SOM prediction accuracy, with a greater importance for mapping SOM in paddy fields than in dry lands; and (4) the local regression results based on the division of paddy fields and dry lands achieved the highest prediction accuracy, with the highest determination coefficient (R2) being 0.653 and lowest root mean square error (RMSE) being 1.144 %. This study proves that different types of arable land have a great impact on the SOM prediction accuracy. Researchers should adopt different strategies to map the SOM contents of paddy fields and dry lands.

Individualized causal mediation analysis with continuous treatment using conditional generative adversarial networks

Traditional methods used in causal mediation analysis with continuous treatment often focus on estimating average causal effects, limiting their applicability in precision medicine. Machine learning techniques have emerged as a powerful approach for precisely estimating individualized causal effects. This paper proposes a novel method called CGAN-ICMA-CT that leverages Conditional Generative Adversarial Networks (CGANs) to infer individualized causal effects with continuous treatment. We thoroughly investigate the convergence properties of CGAN-ICMA-CT and show that the estimated distribution of our inferential conditional generator converges to the true conditional distribution under mild conditions. We conduct numerical experiments to validate the effectiveness of CGAN-ICMA-CT and compare it with four commonly used methods: linear regression, support vector machine regression, decision tree, and random forest regression. The results demonstrate that CGAN-ICMA-CT outperforms these methods regarding accuracy and precision. Furthermore, we apply the CGAN-ICMA-CT model to the real-world Job Corps dataset, showcasing its practical utility. By utilizing CGAN-ICMA-CT, we estimate the individualized causal effects of the Job Corps program on the number of arrests, providing insights into both direct effects and effects mediated through intermediate variables. Our findings confirm the potential of CGAN-ICMA-CT in advancing individualized causal mediation analysis with continuous treatment in precision medicine settings.

Intelligent Crop Management Optimization using Machine Learning Algorithms: A Linear Analytical Approach

This research paper explores Utilizing machine learning techniques in practice, to enhance crop management recommendations. It leverages historical and real-time data, encompassing weather conditions, soil characteristics, growth stages, and past information. A rigorous data prepossessing procedure is employed to generate a well-structured 1.31 lacs data set. Various methodologies are utilized to construct predictive. These models' effectiveness is evaluated using common assessment metrics, such as Recall, accuracy, precision, and F1-score. To enhance the reliability and transparency of the recommendations, ensemble methods such as Naive Bayes, Support vector machines, decision trees, gradient boosting, and random forests, Linear Regression, and k-Nearest Neighbors}, along with interpretative techniques, are incorporated. The results provide insights into different algorithms and predict crucial agricultural activities such as planting, irrigation, fertilization, and pest control. Among all the machine learning models considered, Naive Bayes emerges as the best-performing model, achieving perfect scores of 1.0 in accuracy, precision, recall, and F1-scores. The second-best machine learning model is Random Forests, which follows closely behind with Achieving remarkable results reaching 0.99. These prognostic parameters have a big impact on agricultural productivity and sustainability. In order to improve crop management productivity and resource efficiency, this study promotes the use of data-driven decision-making in agriculture.

Point-of-Care-Ultrasound Quality Assurance Data from Fellow-Performed Exams in a Pediatric Emergency Department: A Descriptive Study.

Point-of-care-ultrasound (POCUS) is increasingly used by pediatric emergency medicine (PEM) fellows, but scant data exists on the accuracy of exam interpretations. Our goal was to determine whether agreement on exam interpretation between quality assurance (QA) faculty (reference standard) and PEM fellows varied by fellowship year or exam type. Retrospective review of fellow-performed POCUS exams between January 2019 and June 2022. Negative binomial (NB) random effects regression was used to account for longitudinal measurement of individual fellow performance across 3 years. Fixed effects were exam type and fellowship year. To assess between- and within-user variability across time, a random intercept and slope were included for each fellow. Exactly 3032 exams, performed by 24 fellows, were included. Raw proportion agreement by fellowship year was high for all exam types (≥88%). From the NB model, there was no statistically significant effect of fellowship year on the mean count of agreement. The relative risk (RR) of agreement for exam types was greatest for cardiac vs other types. The standard deviations for the random intercept and random slope were 0.09 and 0.04, respectively, with a correlation of -0.94. PEM fellows generally interpret exams correctly, with little variation through fellowship, although those who began with more basic skills showed more progress over time. Fellowship year did not influence the likelihood of correct interpretation but there was variation across exam type, with the best agreement for cardiac exams. The extent to which disagreements between fellows and QA faculty represent clinically significant errors requires further study.

Genomic and phenotypic selection indices for decreased birth weight, shortening fattening period, and regulating total weight gain in beef cattle.

To develop genomic and phenotypic indices for beef cattle selection that afford progeny with reduced birth weight and fattening period. The indices should also allow regulating total weight gain during the fattening period and avoid marked increases of inbreeding. Whether addition of constraint on total weight gain to constraints on body weight gain at up to four specific time points during the fattening process was effective for the weight gain until the end of the fattening period was examined in two selection indices with the selection trait being a phenotypic or a genomic breeding value random regression coefficient. Both indices afforded cattle with desired weight gains at specific time points and desired total weight gains. One cycle of index-based selection made it possible to shorten the fattening period two weeks compared with that before selection while maintaining the same final fattening weight as before selection. The impact of constraining total weight gain was smallest when the number of weight gain constraints at specific time points was three or four. However, constraining total weight gain was necessary to avoid poor total weight gain when the number of weight gain constraints at specific time points was only one or two. The developed indices make it possible to regulate total weight gain during the fattening process and to achieve desired weight gains at specific time points. Importantly, the indices bring about genetic improvement without an excessive increase of inbreeding. Thus, we expect that these indices will contribute to sustainable genetic improvement in cattle while maintaining genetic diversity.

Sensing emotional valence and arousal dynamics through automated facial action unit analysis

Information about the concordance between dynamic emotional experiences and objective signals is practically useful. Previous studies have shown that valence dynamics can be estimated by recording electrical activity from the muscles in the brows and cheeks. However, whether facial actions based on video data and analyzed without electrodes can be used for sensing emotion dynamics remains unknown. We investigated this issue by recording video of participants’ faces and obtaining dynamic valence and arousal ratings while they observed emotional films. Action units (AUs) 04 (i.e., brow lowering) and 12 (i.e., lip-corner pulling), detected through an automated analysis of the video data, were negatively and positively correlated with dynamic ratings of subjective valence, respectively. Several other AUs were also correlated with dynamic valence or arousal ratings. Random forest regression modeling, interpreted using the SHapley Additive exPlanation tool, revealed non-linear associations between the AUs and dynamic ratings of valence or arousal. These results suggest that an automated analysis of facial expression video data can be used to estimate dynamic emotional states, which could be applied in various fields including mental health diagnosis, security monitoring, and education.

Credit Card Fraud: Analysis of Feature Extraction Techniques for Ensemble Hidden Markov Model Prediction Approach

In the face of escalating credit card fraud due to the surge in e-commerce activities, effectively distinguishing between legitimate and fraudulent transactions has become increasingly challenging. To address this, various machine learning (ML) techniques have been employed to safeguard cardholders and financial institutions. This article explores the use of the Ensemble Hidden Markov Model (EHMM) combined with two distinct feature extraction methods: principal component analysis (PCA) and a proposed statistical feature set termed MRE, comprising Mean, Relative Amplitude, and Entropy. Both the PCA-EHMM and MRE-EHMM approaches were evaluated using a dataset of European cardholders and demonstrated comparable performance in terms of recall (sensitivity), specificity, precision, and F1-score. Notably, the MRE-EHMM method exhibited significantly reduced computational complexity, making it more suitable for real-time credit card fraud detection. Results also demonstrated that the PCA and MRE approaches perform significantly better when integrated with the EHMM in contrast to the conventional HMM approach. In addition, the proposed MRE-EHMM and PCA-EHMM techniques outperform other classic ML models, including random forest (RF), linear regression (LR), decision trees (DT) and K-nearest neighbour (KNN).

Landscape-scale predictions of future grassland conversion to cropland or development.

Grassland conservation planning often focuses on high-risk landscapes, but many grassland conversion models are not designed to optimize conservation planning because they lack multidimensional risk assessments and are misaligned with ecological and conservation delivery scales. To aid grassland conservation planning, we developed landscape-scale models at relevant scales that predict future (2021-2031) total and proportional loss of unprotected grassland to cropland or development. We developed models for 20 ecoregions across the contiguous United States by relating past conversion (2011-2021) to a suite of covariates in random forest regression models and applying the models to contemporary covariates to predict future loss. Overall, grassland loss models performed well, and explanatory power varied spatially across ecoregions (total loss model: weighted group mean R2=0.89 [range: 0.83-0.96], root mean squared error [RMSE]=9.29ha [range: 2.83-22.77ha]; proportional loss model: weighted group mean R2=0.74 [range: 0.64-0.87], RMSE=0.03 [range: 0.02-0.06]). Amount of crop in the landscape and distance to cities, ethanol plants, and concentrated animal feeding operations had high variable importance in both models. Total grass loss was greater when there were moderate amounts of grass, crop, or development (∼50%) in the landscape. Proportional grass loss was greater when there was less grass (∼<30%) and more crop or development (∼>50%). Some variables had a large effect on only a subset of ecoregions, for example, grass loss was greater when ∼>70% of the landscape was enrolled in the Conservation Reserve Program. Our methods provide a simple and flexible approach for developing risk layers well suited for conservation that can be extended globally. Our conversion models can support conservation planning by enabling prioritization as a function of risk that can be further optimized by incorporating biological value and cost.

Remote sensing for estimating genetic parameters of biomass accumulation and modeling stability of growth curves in alfalfa.

Multi-spectral imaging by unoccupied aerial vehicles provides a non-destructive, high throughput approach to measuring biomass accumulation over successive alfalfa (Medicago sativa L. subsp. sativa) harvests. Information from estimated growth curves can be used to infer harvest biomass and to gain insights into the relationship between growth dynamics and forage biomass stability across cuttings and years. In this study, multi-spectral imaging and several common vegetation indices were used to estimate genetic parameters and model growth of alfalfa cultivars to determine the longitudinal relationship between vegetation indices and forage biomass. Results showed moderate heritability for vegetation indices, with median plot level heritability ranging from 0.11-0.64, across multiple cuttings in three trials planted in Ithaca, NY, and Las Cruces, NM. Genetic correlations between the normalized difference vegetation index and forage biomass were moderate to high across trials, cuttings, and the timing of multi-spectral image capture. To evaluate the relationship between growth parameters and forage biomass stability across cuttings and environmental conditions, random regression modeling approaches were used to estimate the growth parameters of cultivars for each cutting and the variance in growth was compared to the variance in genetic estimates of forage biomass yield across cuttings. These analyses revealed high correspondence between stability in growth parameters and stability of forage yield. The results of this study indicate that vegetation indices are effective at modeling genetic components of biomass accumulation, presenting opportunities for more efficient screening of cultivars and new longitudinal modeling approaches that can provide insights into temporal factors influencing cultivar stability.

Prediction and prevention of concrete chloride penetration: machine learning and MICP techniques

The chloride migration coefficient (CMC) of concrete is crucial for evaluating its durability. This study develops ensemble models to predict the CMC of concrete, addressing the limitations of traditional, labor-intensive laboratory tests. We developed three ensemble models: an inverse variance-based model, an Artificial Neural Network (ANN)-based model, and a tree-based model using the random forest regression algorithm. These models were trained on a dataset comprising 843 concrete mix proportions from existing literature. Results indicate that ensemble models outperform single models such as ANN and Support Vector Regression (SVR) in predicting CMC, with the combined random forest and ANN model showing the highest accuracy. Sensitivity analysis using Shapley Additive Explanations (SHAP) reveals that the CMC is most influenced by the water-to-cement ratio and curing age. Additionally, we designed a graphical user interface (GUI) to facilitate the practical application of our models. This research offers a robust methodology for evaluating concrete durability and potential for extending the prediction to other concrete properties.

Evaluation of Scikit-Learn Machine Learning Algorithms for Improving CMA-WSP v2.0 Solar Radiation Prediction

This article aims to evaluate the performance of solar radiation forecasts produced by CMA-WSP v2.0 (version 2 of the China Meteorological Administration Wind and Solar Energy Prediction System) and to explore the application of machine learning algorithms from the scikit-learn Python library to improve the solar radiation prediction made by the CMA-WSP v2.0. It is found that the performance of the solar radiation forecasting from the CMA-WSP v2.0 is closely related to the weather conditions, with notable diurnal fluctuations. The mean absolute percentage error (MAPE) produced by the CMA-WSP v2.0 is approximately 74% between 11:00 and 13:00. However, the MAPE ranges from 193% to 242% at 07:00–08:00 and 17:00–18:00, which is greater than that observed at other daytime periods. The MAPE is relatively low (high) for both sunny and cloudy (overcast and rainy) conditions, with a high probability of an absolute percentage error below 25% (above 100%). The forecasts tend to underestimate (overestimate) the observed solar radiation in sunny and cloudy (overcast and rainy) conditions. By applying machine learning models (such as linear regression, decision trees, K-nearest neighbors, random forests regression, adaptive boosting, and gradient boosting regression) to revise the solar radiation forecasts, the MAPE produced by the CMA-WSP v2.0 is significantly reduced. The reduction in the MAPE is closely connected to the weather conditions. The models of K-nearest neighbors, random forests regression, and decision trees can reduce the MAPE in all weather conditions. The K-nearest neighbor model exhibits the most optimal performance among these models, particularly in rainy conditions. The random forest regression model demonstrates the second-best performance compared to that of the K-nearest neighbor model. The gradient boosting regression model has been observed to reduce the MAPE of the CMA-WSP v2.0 in all weather conditions except rainy. In contrast, the adaptive boosting (linear regression) model exhibited a diminished capacity to improve the CMA-WSP v2.0 solar radiation prediction, with a slight reduction in MAPE observed only in sunny (sunny and cloudy) conditions. In addition, the input feature selection has a considerable influence on the performance of the machine learning model. The incorporation of the time series data associated with the diurnal variation of solar radiation as an input feature can further improve the model’s performance.

Discovering the spatial heterogeneous constraints of distance on migration from counties to Shenzhen in China

Distance holds significant importance in the decision-making processes of migration flows. Previous studies have predominantly adopted a global perspective to understand distance constraints. However, same spatial heterogeneity exists in distance constraints, and this heterogeneity contributes to understanding of the effect of regional characteristics in human migration. This study answers three questions: Does distance work for all counties in the same way? Why distance works for some counties but does not for others? And what makes distance decay differently among counties? We adopt geographically weighted regression, binary logistics regression, and random forest regression to analyze the migration from counties all over the China to Shenzhen to explore the spatial heterogeneity of distance decay. The results show that distance does not always work, and that demographics and transportation facilities are important determinants of whether distance works. For counties where distance works, distance decay is nonlinearly related to regional development. This non-linear relationship is due to the game between the appeal of intervention opportunities in the nearest new first-tier city, the constraints represented by the socio-economic development of the source county, and the attractiveness of Shenzhen. This further reflects the dynamic interplay between costs and opportunities, give and gain in the decision process of migration.

Study on the Estimation of Leaf Area Index in Rice Based on UAV RGB and Multispectral Data

Leaf area index (LAI) is a key variable for monitoring crop growth. Compared with traditional measurement methods, unmanned aerial vehicle (UAV) remote sensing offers a cost-effective and efficient approach for rapidly obtaining crop LAI. Although there is extensive research on rice LAI estimation, many studies suffer from the limitations of models that are only applicable to specific scenarios with unclear applicability conditions. In this study, we selected commonly used RGB and multispectral (Ms) data sources, which contain three channels of color information and five multi-band information, respectively, combined with five different spatial resolutions of data at intervals of 20–100 m. We evaluated the effectiveness of models using single- and multi-feature variables for LAI estimation in rice. In addition, texture and coverage features other than spectra were introduced to further analyze their effects on the inversion accuracy of the LAI. The results show that the accuracy of the model established with multi-variables under single features is significantly higher than that of the model established with single variables under single features. The best results were obtained using the RFR (random forest regression) model, in which the model’s R2 is 0.675 and RMSE is 0.886 for multi-feature VIs at 40 m. Compared with the analysis results of Ms and RGB data at different heights, the accuracy of Ms data estimation results fluctuates slightly and is less sensitive to spatial resolution, while the accuracy of the results based on RGB data gradually decreases with the increase in height. The estimation accuracies of both Ms and RGB data were improved by adding texture features and coverage features, and their R2 improved by 9.1% and 7.3% on average. The best estimation heights (spatial resolution) of the two data sources were 40 m (2.2 cm) and 20 m (0.4 cm), with R2 of 0.724 and 0.673, and RMSE of 0.810 and 0.881. This study provides an important reference for the estimation of rice LAI based on RGB and Ms data acquired using the UAV platform.

Seasonal dynamics of land surface temperature and urban thermal comfort with land use land cover pattern in semi-arid Indian cities: Insights for sustainable Urban Management

The cities in semi-arid and arid regions generally exhibit distinctive land use land cover (LULC) pattern and seasonal variation in urban thermal comfort (UTC) and land surface temperature (LST). This study examines the seasonal variation in LST and UTC and quantify the importance of LULC pattern in influencing LST in eight Indian semi-arid cities using Landsat 8 datasets. Random forest regression (RFR) model has been applied to quantify the influence of LULC in determining LST. The study shows that bare soil and open spaces in outskirts of selected cities have comparatively high LST than the core of cities. The mean LST varies seasonally from 12.84 °C in Udaipur to 26.18 °C in Jaipur in winter and summer. Analysis of UTC shows that autumn and winter seasons have better UTC than spring and summer seasons. RFR analysis shows a robust correlation (R2 > 0.85) between LULC and LST across all cities. Notably, built-up areas, open land, and vegetation cover exert the greatest influence on LST. Study suggests that targeted LULC planning may effectively address UTC problem in semi-arid cities and enhance sustainability. Study may help in understanding the seasonality thermal environment and assist in mitigating UHI and maintain UTC in these cities.

Effects of season and water quality on community structure of planktonic eukaryotes in the Chaohu Lake Basin.

Analyzing the correlation between planktonic eukaryotic communities (PECs) and aquatic physicochemical parameters (APPs) provides important references for predicting the impact of climate change and human activities on aquatic ecosystems. To assess the influence of seasons and APPs on PEC structures in lakes and rivers, we utilized high-throughput sequencing of the 18S rRNA gene to analyze PEC structures in a lake and seven rivers in the Chaohu Lake Basin and analyzed their correlations with APPs. Our results revealed that PEC structure was significantly affected by season, with the highest α-diversity observed in summer. Furthermore, we identified several APPs, including water temperature, conductivity, dissolved oxygen, pH, phosphate, total phosphorus, trophic level index (TLI), nitrate, ammonia nitrogen, and total nitrogen, that significantly influenced PEC structures. Specifically, we found that Stephanodiscus hantzschii, Simocephalus serrulatus, Cryptomonas sp. CCAC_0109, Pedospumella encystans, Actinochloris sphaerica, Chlamydomonas angulosa, Gonyostomum semen, Skeletonema potamos, Chlamydomonas klinobasis, Pedospumella sp., and Neochlorosarcina negevensis were significantly correlated to TLI, while Limnoithona tetraspina, Theileria sp., and Pseudophyllomitus vesiculosus were significantly correlated to the water quality index (WQI). However, our random forest regression analysis using the top 100 species was unable to accurately predict the WQI and TLI. These results provide valuable data for evaluating the impact of APPs on PEC and for protecting water resource in the Chaohu Lake Basin.

Predicting Arsenic Contamination in Groundwater: A Comparative Analysis of Machine Learning Models in Coastal Floodplains and Inland Basins

Arsenic (As) contamination in groundwater represents a major global health threat, potentially impacting billions of individuals. Elevated As concentrations are found in river floodplains across south and southeast Asia, as well as in the inland basins of China, despite varying sedimentological and hydrogeochemical conditions. The specific mechanisms responsible for these high As levels remain poorly understood, complicating efforts to predict and manage the contamination. Applying hydro-chemical, geological, and soil parameters as explanatory variables, this study employs multiple linear regression (MLIR) and random forest regression (RFR) models to estimate groundwater As concentrations in these regions. Additionally, random forest classification (RFC) and multivariate logistic regression (MLOR) models are applied to predict the probability of As levels exceeding 10 μg/L in the Hetao Basin (China) and Bangladesh. Model validation reveals that RFR explains 80% and 70% of spatial variability of As concentration in the Hetao Basin and Bangladesh, respectively, outperforming MLIR, which accounts for only 35% and 32%. Similarly, RFC outperforms MLOR in predicting high As probability, achieving correct classification rates of 98.70% (Hetao Basin) and 98.25% (Bangladesh) on training datasets, and 82.76% (Hetao Basin) and 91.20% (Bangladesh) on validation datasets. The performance of the MLOR model on the validation set yields accuracy rates of 81.60% and 72.18%, respectively. In the Hetao Basin, Ca2+, redox potential (Eh), Fe, pH, SO42−, and Cl− are key predictors of As contamination, while in Bangladesh, soil organic carbon (SOC), pH, and SO42− are significant predictors. This study underscores the potential of random forest (RF) models as robust tools for predicting groundwater As contamination.

Information and Communications Technology, Water and Sanitation Infrastructures and Economic Development in Sub-Saharan Africa

This study investigated the effect of information and communication technology and water sanitation infrastructures on economic growth and development in sub-Saharan Africa using panel data involving fifteen (15) sub-Saharan Africa nations. Secondary data were obtained from the statistical bulletins of the various countries and World Bank Indicators from 2000-2022. Data obtained were analyzed via descriptive, diagnostic and inferential statistics. The fixed and random effects regression results revealed that information communication technology and water and sanitation infrastructures jointly influence the level of economic growth and development in sub-Saharan Africa. Given the findings, it is recommended that there is the need for governments in sub-Saharan Africa to increase their contributions and supports to ICT infrastructure by way of improving the current state of ICT in sub-Saharan Africa. Furthermore, governments of sub-Saharan African countries should strive towards improving the water and sanitation infrastructure by periodically updating their public health conditions related to clean drinking water and treatment and disposal of human excreta and sewage. Keywords: Information Communication Technology Infrastructure, Water and Sanitation Infrastructure, Economic Growth, Critical Infrastructure, Sub-Saharan Africa

Development and validation of a prognostic model to predict birth weight: individual participant data meta-analysis

ObjectiveTo predict birth weight at various potential gestational ages of delivery based on data routinely available at the first antenatal visit.DesignIndividual participant data meta-analysis.Data sourcesIndividual participant data of four cohorts...

Accurate Prediction of Rat Acute Oral Toxicity and Reference Dose for Thousands of Polycyclic Aromatic Hydrocarbon Derivatives Based on Chemometric QSAR and Machine Learning.

Acute oral toxicity is currently not available for most polycyclic aromatic hydrocarbons (PAHs), especially their derivatives, because it is cost-prohibitive to experimentally determine all of them. Here, quantitative structure-activity relationship (QSAR) models using machine learning (ML) for predicting the toxicity of PAH derivatives were developed, based on oral toxicity data points of 788 individual substances of rats. Both the individual ML algorithm gradient boosting regression trees (GBRT) and the stacking ML algorithm (extreme gradient boosting + GBRT + random forest regression) provided the best prediction results with satisfactory determination coefficients for both cross-validation and the test set. It was found that those PAH derivatives with fewer polar hydrogens, more large-sized atoms, more branches, and lower polarizability have higher toxicity. Software based on the optimal ML-QSAR model was successfully developed to expand the application potential of the developed model, obtaining reliable prediction of pLD50 values and reference doses for 6893 external PAH derivatives. Among these chemicals, 472 were identified as moderately or highly toxic; 10 out of them had clear environment detection or use records. The findings provide valuable insights into the toxicity of PAHs and their derivatives, offering a standard platform for effectively evaluating chemical toxicity using ML-QSAR models.

Momentum prediction models of tennis match based on CatBoost regression and random forest algorithms

As we all know, momentum plays a crucial role in ball game. Based on the 2023 Wimbledon final data, this paper investigated momentum in tennis. Firstly, we initially trained a decision tree regression model on reprocessed data for prediction, and established the CBRF model based on CatBoost regression and random forest regression models to obtain prediction data. Secondly, significant non-zero autocorrelation coefficients were found, confirming the correlation between momentum and success. Thirdly, Based on these key factors, we proposed winning strategies for the players, conducted predictive analyses for six specific time intervals of the game. At last, by implementing these models to women’s matches, championships, matches on different surfaces, the results demonstrated that the models have effective generalization ability.