Regression Accuracy Research Articles

Multi-scale spatiotemporal wetland loss and its critical influencing factors in China determined using innovative grid-based GWR

Wetlands serve as a critical habitat for many plants and animals. However, with urban expansion and other processes, the wetlands in China are lost at an alarming rate in recent decades. This study quantifies the pace of wetland loss in the past two decades and investigates the hidden mechanisms of wetland loss via a proposed innovative grid-based geographically weighted regression computing unified device architecture (Grid-GWR-CUDA) method. Also assessed in this study is the relative importance of four natural factors (precipitation, digital elevation model (DEM), slope, evaporation), and four human related factors (distance to city, distance to roads, population, and gross domestic product (GDP)) in wetland loss at both the national and regional levels. The importance of some critical factors is compared across different regions. The results indicated that GDP, population size and DEM are the three most significant factors impacting on the rate of wetland loss at the national level. They are still the most influential factors for most regions at the sub-national level while precipitation is an important factor only in the eastern regions. Therefore, the main causes of wetland loss in China are attributed to human socioeconomic activities. Compared with the traditional geographically weighted regression (GWR) method, the proposed Grid-GWR-CUDA method is innovative enough to handle large data volumes efficiently while retaining all data samples at a fine (e.g., 30 m) spatial resolution because the computational efficiency is significantly improved via graphics processing unit (GPU) acceleration. It improves the accuracy of regression by several folds over traditional ordinary least square and GWR. It has the potential to predict future wetland loss if supplied with the latest land cover data.

Tomato recognition and location algorithm based on improved YOLOv5

In order to meet the requirements of target detection and location for intelligent tomato picking, a recognition and location algorithm based on improved YOLOv5s is proposed in this paper. The CARAFE module structure was used to optimize the upsampling algorithm of YOLOv5s, which increased the receptive field of the network while maintaining the lightweight. EIoU and Quality Focal Loss were introduced to optimize the loss function of the network, which solved the problem of reduced accuracy caused by uneven samples, and at the same time accelerated the training convergence speed and improved the regression accuracy. The improved model is denoted as YOLOv5s-CQE. Compared with YOLOv5s, the mAP_0.5 and mAP_0.5:0.95 of YOLOv5s-CQE was increased by 1.67 and 3.43 percentages points, respectively. The recognition accuracy of YOLOv5s-CQE model in the test set was 99.77%, which had increased by 2.40 percentages points compared with before improvement, and was 4.10, 6.47 and 3.03 percentages points higher than that of the lightweight networks YOLOv4-tiny, YOLOv5-Lite-e and YOLOv5-Lite-s, respectively. Compared with YOLOv7 and Faster RCNN, the recognition accuracy rate was increased by 1.69, 3.97 percentages points respectively. In order to improve the accuracy of positioning, distortion removal and ROI clipping were carried out on the obtained images. The accuracy of location was tested by laboratory positioning test and field picking test. The results showed the total average errors decreased by 6.65 mm compared with those before distortion removal, and the field positioning accuracy was improved by 6.67 percentages points. The experimental results showed that the algorithm in this paper had the advantages of high precision, fast detection speed and strong robustness, which provided a theoretical basis for intelligent tomato picking.

Object Detection for UAV Aerial Scenarios Based on Vectorized IOU.

Object detection in unmanned aerial vehicle (UAV) images is an extremely challenging task and involves problems such as multi-scale objects, a high proportion of small objects, and high overlap between objects. To address these issues, first, we design a Vectorized Intersection Over Union (VIOU) loss based on YOLOv5s. This loss uses the width and height of the bounding box as a vector to construct a cosine function that corresponds to the size of the box and the aspect ratio and directly compares the center point value of the box to improve the accuracy of the bounding box regression. Second, we propose a Progressive Feature Fusion Network (PFFN) that addresses the issue of insufficient semantic extraction of shallow features by Panet. This allows each node of the network to fuse semantic information from deep layers with features from the current layer, thus significantly improving the detection ability of small objects in multi-scale scenes. Finally, we propose an Asymmetric Decoupled (AD) head, which separates the classification network from the regression network and improves the classification and regression capabilities of the network. Our proposed method results in significant improvements on two benchmark datasets compared to YOLOv5s. On the VisDrone 2019 dataset, the performance increased by 9.7% from 34.9% to 44.6%, and on the DOTA dataset, the performance increased by 2.1%.

Study of the Profile Distribution of the Diffuse Attenuation Coefficient and Secchi Disk Depth in the Northwestern South China Sea

The diffuse attenuation coefficient for the downwelling irradiance Kdz,λ is a critical parameter in terms of the optical properties of the ocean. In the northwestern South China Sea, there are complex physical processes, and the accurate estimation of Kdz,λ in the northwestern South China Sea is critical for the study and application of the underwater light field and water constituents. In this study, using Hydrolight 6.0 (HL60) software, Kdz,λ was simulated based on the inherent optical properties (IOPs) and chlorophyll a concentration dataset in the northwestern South China Sea. The simulations were in good agreement with the results calculated by the model of Lee (2005), and the spectral characteristics of Kdz,λ were consistent with several oceanic types according to Jerlov’s classification. The horizontal and vertical distribution characteristics of Kd(z,490) were studied in the two typical upwelling areas of eastern Hainan Island and eastern Vietnam. Kd(z,490) in eastern Hainan Island exhibited an overall decreasing trend from west to east at the same depth, while the vertical depth of the maximum value of Kd(z,490) in eastern Hainan Island was found to increase from west to east, which was significantly associated with the distribution trend of the temperature and salinity. Kd(z,490) in eastern Vietnam exhibited unique horizontal and vertical distribution characteristics due to upwelling, with a low temperature and high salinity. A satisfactory linear relationship between Kd(z,490) and Kdz,λ was found from 420 to 580 nm with R2 > 0.76, root mean square (RMSE) ≤ 0.010 m−1, and mean absolute percentage error (MAPE) < 9%, and this result indicated that Kdz,λ from 420~580 nm could be estimated with Kd(z,490). The regression accuracy sharply decreased after 580 nm, indicating that Kdz,λ estimation based on Kd(z,490) can be more suitably achieved from 420~580 nm and becomes inaccurate after 580 nm. Based on the simulations, an empirical relationship for ZSDD estimation involving Kd,10%490 was developed, and ZSDD in the northwestern South China Sea was calculated, with a range of 5–23 m and a suitable agreement with ZSDD obtained via the method of Lee (2018).

An efficient hybrid LSTM-ANN joint classification-regression model for PPG based blood pressure monitoring

This paper investigates the importance of classification in optimizing the estimation accuracy of blood pressure (BP) using photoplethysmography (PPG) signal features, with the aim of balancing efficiency and performance. Accordingly, the experimental design of this study was employed by constructing various artificial neural network (ANN) and long short-term memory (LSTM) topologies such as conventional single-stage models and classification-regression schemes including two-stage models, joint learning models, and the hybrid LSTM-ANN model. All models were tested on 40 subjects from the Physionet’s MIMIC II database. With only 4 features, the proposed hybrid LSTM-ANN model achieved the best MAE ± SD of 3.39 ± 5.47 and 1.79 ± 3.72 mmHg for SBP and DBP, respectively. Furthermore, the single-stage and the two-stage ANN-based models were selected to be embedded in STM32 Microcontroller Unit (MCU) for real-time predictions of BP. The two-stage classification-regression ANN model showed superior performance and robustness in real-time testing on six subjects, achieved an MAE ± SD of 1.41 ± 1.29 mmHg, resulting in an 83.5% reduction of MAE compared with the traditional single-stage ANN model. Overall, the data presented in this work substantiate the significant role of classification in improving regression accuracy in both ANN and LSTM-based models. Compared to other state-of-the-art models, the hybrid LSTM-ANN with a joint classification-regression approach was found to be the optimal architecture to achieve the best performance and a trade-off between memory usage, complexity, and latency with acceptable BP prediction accuracy.

Forecasting the eddy current loss of a large turbo generator using hybrid ensemble Gaussian process regression

For the issue that the sample space of wedge winding eddy current losses of large generator does not obey Gaussian distribution, a hybrid ensemble Gaussian process regression (HEGPR) model is proposed in this paper. The HEGPR contains three layers. First, four tree regression models (XGBoost, CatBoost, LGBM and NGBoost) are built. Then, the output of the first layer is taken as the input of multiple Gaussian regression models, so that the input samples of the second layer obey Gaussian distribution, which can effectively improve the generalization ability of Gaussian process regression. The results show that the root mean squared error (RMSE) is 0.0282 and the goodness of fit (R2) is 0.9973. The model has good prediction performance for the eddy current loss of large turbo generator. Compared with kinds of Gaussian process models and traditional ensemble learning models, the prediction accuracy of this model is higher, and it is more suitable for forecasting eddy current loss of the large generator. HEGPR model can effectively solve the problem of insufficient regression accuracy of Gaussian process when sample space does not obey Gaussian distribution.

Forecasting accuracy of machine learning and linear regression: evidence from the secondary CAT bond market

AbstractThe main challenge in empirical asset pricing is forecasting the future value of assets traded in financial markets with a high level of accuracy. Because machine learning methods can model relationships between explanatory and dependent variables based on complex, non-linear, and/or non-parametric structures, it is not surprising that machine learning approaches have shown promising forecasting results and significantly outperform traditional regression methods. Corresponding results were achieved for CAT bond premia forecasts in the primary market. However, since secondary market data sets have a panel data structure, it is unclear whether the results of primary market studies can be applied to the secondary market. Against this background, this study aims to build the first out-of-sample forecasting model for CAT bond premia in the secondary market, comparing different modeling approaches. We apply random forest and neural networks as representatives of machine learning methods and linear regression based on a comprehensive data set of CAT bond issues and across various forecasting settings and show that random forest forecasts are significantly more precise. Because the lack of transparency of machine learning methods may limit their applicability, especially for institutional investors, we show ways to identify important variables in the context of random forest price forecasting.

Prediction of line heating deformation for sheet based on SSA-ELM algorithm

Line heating is highly susceptible to plastic deformation leading to structural changes, which seriously affects shipbuilding efficiency. It is extremely important to obtain the precise line heating deformation law. There is much research on plate deformation prediction, but there is still no mature, accurate and fast prediction method. Based on this, it is necessary to research the line heating deformation prediction method. ELM has the advantages of fast learning speed, good generalization performance and strong function approximation ability. To address the poor stability of the regression model due to the randomly generated weights and thresholds of ELM, a combined prediction model (SSA-ELM) based on a squirrel search algorithm (SSA) optimized limit learning machine is proposed. The results show that the relative errors of transverse shrinkage and angular deformation of SSA-ELM are improved by 4.7% and 3.9%. The SSA-ELM prediction model has good regression accuracy and generalization ability. It gives more accurate prediction results in terms of line heating deformation prediction.

Real-Time Intelligent Prediction Method of Cable’s Fundamental Frequency for Intelligent Maintenance of Cable-Stayed Bridges

Cable’s fundamental frequency (CFF) is an important characteristic of the working state of long-span cable-stayed bridges. The change in the bridge’s temperature field will influence CFF by altering the cable’s tension and the cables’ sags. An accurate regression model between the temperature-induced variation of CFF and the real-time changing temperature field should be established. Then, the reference value of the temperature-induced variation of CFF can be obtained after inputting the real-time temperature data. In this study, an intelligent real-time prediction model for CFF is proposed based on the full-bridge temperature field, including the average temperature of the main beam, the vertical temperature difference of the main beam, and the temperature of the cable tower. Besides, a machine learning method named the long short-term memory (LSTM) network is exploited to ensure the nonlinear fitting performance of the model, and a paradigm for optimal hyperparameter selection and training strategy selection is provided. To verify the superiority of the LSTM-based model, the output accuracy of the linear regression, BP network, and LSTM network was tested and compared using the monitoring data collected from cable sensors in the main span and side span, which provides an important basis for the intelligent maintenance and sustainable operation of the bridge cables.

Risk Prediction for the Development of Hyperuricemia: Model Development Using an Occupational Health Examination Dataset

Hyperuricemia has become the second most common metabolic disease in China after diabetes, and the disease burden is not optimistic. We used the method of retrospective cohort studies, a baseline survey completed from January to September 2017, and a follow-up survey completed from March to September 2019. A group of 2992 steelworkers was used as the study population. Three models of Logistic regression, CNN, and XG Boost were established to predict HUA incidence in steelworkers, respectively. The predictive effects of the three models were evaluated in terms of discrimination, calibration, and clinical applicability. The training set results show that the accuracy of the Logistic regression, CNN, and XG Boost models was 84.4, 86.8, and 86.6, sensitivity was 68.4, 72.3, and 81.5, specificity was 82.0, 85.7, and 86.8, the area under the ROC curve was 0.734, 0.724, and 0.806, and Brier score was 0.121, 0.194, and 0.095, respectively. The XG Boost model effect evaluation index was better than the other two models, and similar results were obtained in the validation set. In terms of clinical applicability, the XG Boost model had higher clinical applicability than the Logistic regression and CNN models. The prediction effect of the XG Boost model was better than the CNN and Logistic regression models and was suitable for the prediction of HUA onset risk in steelworkers.

Analysis and Comparison for Prediction of Diabetic among Pregnant Women using Innovative K-Nearest Neighbor Algorithm over Logistic Regression with Improved Accuracy

Aim: In comparison to the K-Nearest Neighbor (KNN) algorithm, Logistic Regression (LR) was used in machine learning algorithms for the prediction of diabetes among pregnant women to get better accuracy, PEARCEsensitivity, and precision. Materials and methods: To verify the usefulness of the technique, researchers collected data sets from free available data sets such as the Pima Indian dataset from the UCI website to examine diabetes among pregnant women. There are two groups in this study: K-Nearest Neighbor (N=20) and Logistic Regression (N=20), each having a sample size of 40. A pre-test power of 80%, a threshold of 0.05, and a confidence interval of 95% are used in the sample size calculation. Results: The accuracy, sensitivity, and precision of algorithms are used to evaluate their performance. K-Nearest Neighbor (KNN) accuracy rate is 72.44 percent, but Linear Regression (LR) accuracy is 76.67%. The sensitivity rate for K-Nearest Neighbor is 74.42 percent, while the sensitivity rate for Linear Regression (LR) is 76.16 percent. The precision rate for K-Nearest Neighbor (KNN) is 73.75percent, whereas the precision rate for Linear Regression (LR) is 81.87 percent. The accuracy rate is significantly different P=0.366 (P>0.05). Conclusion: When compared to the Innovative K-Nearest Neighbor algorithm, the Logistic Regression algorithm predicts better classification in discovering the accuracy, sensitivity, and precision for accessing the rate for prediction of diabetes among pregnant women.

Analysis of Soft Skills and Job Level with Data Science: A Case for Graduates of a Private University

This study shows the significant features predicting graduates’ job levels, particularly high-level positions. Moreover, it shows that data science methodologies can accurately predict graduate outcomes. The dataset used to analyze graduate outcomes was derived from a private educational institution survey. The original dataset contains information on 17,898 graduates and approximately 148 features. Three machine learning algorithms, namely, decision trees, random forest, and gradient boosting, were used for data analysis. These three machine learning models were compared with ordinal regression. The results indicate that gradient boosting is the best predictive model, which is 6% higher than the ordinal regression accuracy. The SHapley Additive exPlanations (SHAP), a novel methodology to extract the significant features of different machine learning algorithms, was then used to extract the most important features of the gradient boosting model. Current salary is the most important feature in predicting job levels. Interestingly, graduates who realized the importance of communication skills and teamwork to be good leaders also had higher job positions. Finally, general relevant features to predict job levels include the number of people directly in charge, company size, seniority, and satisfaction with income.

Evaluating parallelized support vector regression and nearest neighbor regression with different input variations for estimating daily global solar radiation of the humid subtropical region in China

AbstractIndirectly estimating global solar radiation is strongly nonlinear and needs to be addressed by machine learning. Sequentially developing a machine learning model can be very time consuming. Moreover, whether and how the exogenous meteorological, geographical and temporal variables affect regression accuracy still has not been well understood. This paper evaluates parallelized support vector regression (SVR) and nearest neighbor regression (NNR) models for estimating daily global solar radiation of the humid subtropical region in China using existing Python libraries on a multi-core central processing unit (CPU) and a graphical processing unit (GPU). Seven input variations are studied. Two variations are commonly adopted in literature, four variations contain meteorological, geographical and/or temporal features with bounded Pearson correlation coefficients (PCCs), and the other variation simply include all the available features. Experimental results demonstrate that: SVR and NNR are equally powerful for nonlinear regression, and the variation comprising features with absolute PCCs no less than 0.3 (i.e. just all the meteorological features) is able to achieve most accurate estimation; the GPU-parallelized SVR model can accelerate parameter calibration and prediction; compared with the CPU-parallelized and GPU-parallelized SVR models, the GPU-parallelized NNR model is much more efficient and rather more scalable with the increment of the number of data samples; and the CPU-parallelized NNR model consumes quite less parameter calibration time than the GPU-parallelized NNR model, owing to different methods adopted for determining distances and significant time wasted by the GPU-parallelized NNR model on repeatedly calculating required information during cross-validation.

Machine learning algorithm for the structural design of MEMS resonators

MEMS resonators have become core devices in many fields, their geometric designs can profoundly affect performance. However, the theoretical modeling of MEMS resonators with complex structures is becoming more and more difficult, and the time consumption of using numerical simulation methods to solve the accurate analytical solution of performance is also increasing. In this paper, we report a working mode identification and a machine learning algorithm, which dramatically shorten the MEMS design cycle by constructing datasets and predicting physical properties with high speed and high accuracy. As an example, we apply the algorithms to the performance prediction of flower-like disk resonator. The typical structural parameters of MEMS resonators are used as the input layer of the neural network, and performance generated by finite element analysis methods are used as the output layer. For 50,000 modal shapes with 5000 different structural parameters, the accuracy of the working mode identification algorithm proposed in this paper to identify elliptical modes is 100%. After sufficient training, the obtained neural network calculators can predict the resonant frequency, thermoelastic quality factor, mechanical sensitivity and mechanical thermal noise of the MEMS resonator. Compared with traditional numerical simulation methods, the identification of resonant frequency and thermoelastic quality factor is 11,341 times faster, the identification of mechanical sensitivity and mechanical thermal noise is 1813 times faster, and the prediction regression accuracy is all greater than 96%. This high-speed and high-accuracy performance prediction method can effectively improve the design efficiency of MEMS resonators with complex structures, providing a promising tool for enhancing MEMS resonator performance.

Assessing maize lodging severity using multitemporal UAV-based digital images

Lodging is a common agricultural disaster in the maize growing period that damages photosynthesis and even affects yield, quality and mechanical harvesting. The purpose of this study was to assess maize lodging severity using the changes in variables extracted from unmanned aerial vehicle (UAV) orthophoto images before and after maize lodging. Maize lodging severity included non-lodging, mild lodging, moderate lodging and severe lodging. We extracted the red, green, blue bands, hue, saturation and value (HSV) colour and texture information before and after maize lodging. Variables sensitive to maize lodging severity were selected using correlation analysis and the Boruta algorithm. Two linear regression and three machine learning methods were used to classify maize lodging severity. From the results of the correlation analysis, 27 variables were significantly related to maize lodging severity at the 0.01 level. Saturation and value were highly correlated with maize lodging severity, with r values of − 0.92 and 0.81, respectively. In addition, 22 variables sensitive to maize lodging severity were selected by the Boruta algorithm. Among the five estimation models built by using sensitive variables, each estimation model performed well, with R2 higher than 0.6. For the linear regression models, the estimation accuracy of the multiple linear regression (MLR) model (R2 = 0.89 and RMSE = 0.84 for the training set; R2 = 0.9 and RMSE = 0.35 for the testing set) was higher than that of the partial least squares regression (PLSR) model. For nonlinear regression models, both the R2 of the training set and testing set for each model were above 0.8. The random forest (RF) estimation model had the highest accuracy among the five models, with an overall accuracy of 89.47% and Kappa coefficient of 0.84. This study indicated that UAV digital images could be used to obtain lodging information over a large number of maize breeding plots, which may be helpful to assist breeders in quickly picking out lodging-resistant maize varieties.

Correlated Bayesian Co-Training for Virtual Metrology

A rising challenge in manufacturing data analysis is training robust regression models using limited labeled data. In this work, we investigate a semi-supervised regression scenario, where a manufacturing process operates on multiple mutually correlated states. We exploit this inter-state correlation to improve regression accuracy by developing a novel co-training method, namely Correlated Bayesian Co-training (CBCT). CBCT adopts a block Sparse Bayesian Learning framework to enhance multiple individual regression models which share the same support. Additionally, CBCT casts a unified prior distribution on both the coefficient magnitude and the inter-state correlation. The model parameters are estimated using maximum-a-posteriori estimation (MAP), while hyper-parameters are estimated using the expectation-maximization (EM) algorithm. Experimental results from two industrial examples shows that CBCT successfully leverages inter-state correlation to reduce the modeling error by up to 79.40%, compared to other conventional approaches. This suggests that CBCT is of great value to multi-state manufacturing applications.

Breast Cancer Classification Using Outlier Detection and Variance Inflation Factor

In terms of malignant tumors, breast cancer is one of the most prevalent. Breast cancer is a form of cancer that develops in the breast tissue when the surrounding, healthy breast tissue is overtaken by the uncontrollably growing cells in the breast tissue. Several features or patient conditions can be used in a machine learning approach to predict breast cancer. Machine learning will be utilized in these situations to determine if the cancer is malignant or benign. The Wisconsin Breast Cancer (Diagnostic) Data Set, which contains 32 characteristics and 569 collected data, was the dataset used in this research.. Feature selection in this study is done by eliminating outliers using the upper and lower quartile of each feature then feature selection is also carried out on features that have features that have a high variance inflation factor. The machine learning methods used in this research are Logistic Regression, Random Forest, KNN, SVC, XG Boost, Gradient Boosting, and Ridge Classifier. The selection of this method is based on the target that will be predicted by 2 labels, namely benign cancer, and malignant cancer. The result obtained is that the selection of features using the variance inflation factor increases the accuracy of the previous Logistic Regression and Random Forest methods from 98.25% to 99.12%. The method that has the highest level of accuracy is the Logistic Regression and Random Forest methods which have a value of 99.12%. The next research will be developed by trying other optimization techniques for hyperparameter tuning.

Using Pre-Fire High Point Cloud Density LiDAR Data to Predict Fire Severity in Central Portugal

The wall-to-wall prediction of fuel structural characteristics conducive to high fire severity is essential to provide integrated insights for implementing pre-fire management strategies designed to mitigate the most harmful ecological effects of fire in fire-prone plant communities. Here, we evaluate the potential of high point cloud density LiDAR data from the Portuguese áGiLTerFoRus project to characterize pre-fire surface and canopy fuel structure and predict wildfire severity. The study area corresponds to a pilot LiDAR flight area of around 21,000 ha in central Portugal intersected by a mixed-severity wildfire that occurred one month after the LiDAR survey. Fire severity was assessed through the differenced Normalized Burn Ratio (dNBR) index computed from pre- and post-fire Sentinel-2A Level 2A scenes. In addition to continuous data, fire severity was also categorized (low or high) using appropriate dNBR thresholds for the plant communities in the study area. We computed several metrics related to the pre-fire distribution of surface and canopy fuels strata with a point cloud mean density of 10.9 m−2. The Random Forest (RF) algorithm was used to evaluate the capacity of the set of pre-fire LiDAR metrics to predict continuous and categorized fire severity. The accuracy of RF regression and classification model for continuous and categorized fire severity data, respectively, was remarkably high (pseudo-R2 = 0.57 and overall accuracy = 81%) considering that we only focused on variables related to fuel structure and loading. The pre-fire fuel metrics with the highest contribution to RF models were proxies for horizontal fuel continuity (fractional cover metric) and the distribution of fuel loads and canopy openness up to a 10 m height (density metrics), indicating increased fire severity with higher surface fuel load and higher horizontal and vertical fuel continuity. Results evidence that the technical specifications of LiDAR acquisitions framed within the áGiLTerFoRus project enable accurate fire severity predictions through point cloud data with high density.

Assessment of soil CO2 and NO fluxes in a semi-arid region using machine learning approaches

Agricultural lands are sources and sinks of greenhouse gases (GHGs). The identification of the main drivers affecting GHGs is crucial for planning sustainable agronomic practices and mitigating global warming potential. The main aim of this research was to evaluate the impact of environmental drivers (soil temperature and water-filled pore space, WFPS) and crop residue rates on CO2, NO, and NOx fluxes under conventional tillage (CT) and no-tillage (NT) systems. The accuracy of Random Forest Regression (RFR), Multiple Adaptive Regression Splines (MARS), and General Linear Models (GLM) in predicting CO2, NO, and NOx fluxes were also assessed.In both CT and NT systems, CO2, NO, and NOx fluxes decreased with increasing WFPS. Increasing temperature resulted in higher CO2 emissions and lower NO and NOx emissions. Higher residue rates resulted in significant increases in CO2 emission, whereas the NO and NOx emissions increased by decreasing the ratio of residue. For CO2 prediction, the RFR provided the largest R2 with the observed data. For NO–N and NOx-N prediction, RFR was the most efficient algorithm, but NO–N can be predicted with better accuracy. The output of this research highlights the importance of agronomic practices for climate mitigation, along with the possibility of using RFR to predict GHGs fluxes.

Prediction Research on Irregularly Cavitied Components Volume Based on Gray Correlation and PSO-SVM

The use of a micro-compressed air-volume-detection method to detect the volume of irregularly cavitied components has the characteristics of multi-variable coupling and nonlinearity. To solve this problem, a volume-prediction model of irregularly cavitied components based on gray correlation and a particle-swarm-optimization support-vector machine is proposed. In this paper, the gray-correlation method was used to extract the detection parameters that have the greatest correlation with the cavity volume. On the basis of the obtained detection parameters, the SVM algorithm was used to build an irregularly cavitied components volume-prediction model. During model training, since the regression accuracy and generalization performance of the SVM model depend on the proper setting of its two parameters (the penalty-parameter C and the kernel-parameter σ), and especially on the interaction of the parameters, this paper presents an optimal-selection approach towards the SVM parameters, based on the particle-swarm-optimization (PSO) algorithm. Experiments showed that the prediction model can better predict the volume of irregularly cavitied components, and the prediction accuracy was high, which played a guiding role in intellectual nondestructive testing of the volume of the irregularly cavitied components.