Support Vector Regression Algorithm Research Articles

Thin-film sensors for data-driven concentricity prediction in cup backward extrusion

The manufacture of precise metal components by cold forging poses serious challenges to the process reliability under unstable process conditions. The detection of geometrical imperfections, such as concentricity deviations, is necessary for the operator to adjust the forging tool rack properly. In this paper, a novel piezoelectric thin-film sensor disc is introduced to detect such concentricity deviations based on the measurement of eccentric load, that is arising from elastic punch deformation. Experimental results showed, that the concentricity deviation of the produced parts efficiently can be predicted by processing measured force data using a support vector regression algorithm.

Estimating Remaining Useful Life of Cutting Tools in Machining Using an Extended Kalman Filter

Monitoring the condition of the cutting tool and forecasting the evolution of its wear during machining are vital to ensure workpiece quality and the safety of machine elements. The accurate prediction of tool wear is crucial to advance predictive maintenance in machining. This article presents a methodology to estimate the remaining useful life (RUL) of cutting tools. The methodology involves extracting features from vibration signals using discrete wavelet transform (DWT) and computing the concomitant indicators of tool health by applying distance metrics to the features. These indicators have a high correlation with tool wear measurements. The RUL is estimated by fusing these indicators in a support vector regression (SVR) algorithm. The SVR outputs then serve as inputs to an extended Kalman filter (EKF) that employs a rectified linear activation unit (ReLU) function for state evolution, enabling real-time estimation of the RUL. The proposed methodology is demonstrated on the IEEE PHM 2010 dataset, showcasing its reliability and effectiveness in accurately estimating the RUL for cutting tools.

Assessment of the Support Vector Regression and Random Forest Algorithms in the Bias Correction Process on Temperatures

Climate information can be obtained from General circulation models (GCMs). However, this model has poor resolution, so it is necessary to do bias correction to overcome this problem. This study carried out a bias correction process using the Support Vector Regression (SVR) and Random Forest (RF) approaches. Bias correction is carried out for temperature in Indonesia using the BNU-ESM and MERRA-2 climate models, which act as observational data. The results show that the RF method (RMSE: 0.334; Correlation: 0.694; Standard Deviation: 0.582) is better than SVR (RMSE: 0.341; Correlation: 0.675; Standard Deviation: 0.588) in performing bias correction.

Comparative Study of Iconnet Jabodetabek and Banten Using Linear Regression and Support Vector Regression

PT PLN Indonesia Comnets Plus has little information regarding future customer growth, making it difficult to take steps to meet customer needs. This research aims to predict customer growth in the future based on quantitative data from the previous year, where the output provided produces data in the form of numbers that are analyzed using statistical methods. The hope is to provide information to maximize customer growth with the minimum area or bandwidth used by the company. This research uses linear regression and support vector regression (SVR) algorithms using a company secondary dataset of 252 data points with 5 attributes. Data was collected during the last one-year period, from January to December 2021. The results of the research show that predictions using both algorithms have increased, customer growth when viewed from the number of customers, bandwidth, and regional data has increased significantly. This can be seen from the value of the number of customers, which continues to increase, while the highest number of customers falls in December, the most requested bandwidth is 20 mbps, and the largest customer area is in the Depok area. The results of the research show that the SVR algorithm is superior in terms of mean absolute percentage error (MAPE): 0.02% MAPE, 0.10 MAE, and 0.99 RMSE, while for linear regression, the MAPE values were 36.28%, MAE 201, and RMSE 0.80.

Wholesale Inventory Management Optimization: Methodological Approach with XGBoost, SVR, and Random Forest Algorithms

This research aims to optimize wholesale inventory management at PT Primafood International Pasir Putih 2 by implementing leading algorithms, namely XGBoost, Support Vector Regression (SVR), and Random Forest. In the wholesale industry, effective inventory management plays a crucial role in maintaining smooth production processes and enhancing company profitability. Despite the acknowledged benefits of inventory management, there are aspects that remain not fully disclosed, particularly concerning demand uncertainty and market fluctuations. This study addresses these gaps by exploring the potential of these three algorithms. Experimental methods with a quantitative approach were employed to shape and prepare the dataset. The analysis and predictions' results using XGBoost, SVR, and Random Forest were evaluated using metrics such as Mean Squared Error (MSE), F1-Score, and Accuracy. The evaluation indicates that XGBoost and SVR exhibit optimal performance with low MSE values of 7714.446 and 119.315, high F1-Scores (0.92), and good accuracy levels (0.86 and 0.85), respectively. While Random Forest shows a higher MSE, it still delivers solid performance with an F1-Score of 0.89 and an accuracy rate of 0.81. These findings suggest that all three algorithms can be considered to enhance inventory management performance at PT Primafood International Pasir Putih 2, with significant potential benefits for overall industry development. This research provides valuable insights for decision-making at the business and industrial levels, highlighting the effectiveness of each algorithm in the context of predicting stock level.

Research on the Development of China’s Sports Tourism Industry Based on the Background of Big Data Era

Abstract In this paper, we use the support vector regression algorithm to minimize the loss function between the real value of sports tourism samples and the predicted value of the model, deal with the complex nonlinear problems through the MLP neural network model, and design a sports tourism development model based on the machine learning model. Focusing on the use of a support vector regression algorithm and MLP neural network model to predict the tourist flow, the panel threshold regression model is used to test whether the variables have a threshold effect on each other and the relationship between the sports industry and tourism development is deeply explored. Based on this basis, the current situation and development of sports tourism were analyzed using data from Province S and 29 provinces in China. The results show that there is a threshold effect on the influence of the Chinese sports industry on regional tourism development, and the estimated value of the threshold is 1.8293. The level of the sports industry in Chinese provinces can be divided into less developed regions of the sports industry ≤1.8293 and developed regions of the sports industry >1.8293. In the first model, which only includes the value of the level of the sports industry in the current period, the relationship between the sports industry and regional tourism development presents a “ λ Λ” shape.

Estimation of compressive strength for spiral stirrup-confined circular concrete column using optimized machine learning with interpretable techniques

The compressive strength (CS) of a concrete column confined with spiral stirrups is an important indicator for assessing the safety and stability of concrete structures. However, achieving accurate CS estimation for confined concrete remains challenging due to the complex confinement mechanism provided by spiral stirrups. In this study, three robust machine learning (ML) algorithms—support vector regression (SVR), random forest (RF) and extreme gradient boosting (XGBoost)—are employed to predict the CS value of the spiral stirrup-confined circular concrete columns. The hyperparameters of the ML models undergo fine-tuning via Bayesian optimization with 10-fold cross-validation, and the optimized ML models are evaluated for their predictive capabilities. Results show that compared to SVR and RF, XGBoost exhibits more stable generalization performance, achieving an average coefficient of determination (R 2) of 0.944 for the 10-fold cross-validation, and demonstrates superior accuracy on the testing dataset with an R 2 value of 0.967. To provide insights into the relationship between input features and the output CS value, Individual Conditional Exception (ICE) plots, one/two-dimensional Partial Dependence Plots (PDPs), and Shapley Additive Explanation (SHAP) techniques are utilized to interpret the optimized XGBoost model. Additionally, a friendly online graphical user interface (GUI) has been specially developed based on the optimized XGBoost model to facilitate convenient CS estimation for spiral stirrup-confined circular concrete column.

Investigation on the management of college performance pay with linear programming and mathematical statistics

The performance-related pay system in colleges and universities is one of the hot spots in the reform of the wage system in colleges and universities in recent years. It is an important measure to improve the quality of education and teaching in colleges and universities and the construction of talent teams. However, there are many problems in the performance-related pay system of colleges and universities, such as unreasonable evaluation indicators, unfair weight distribution, etc., which makes it difficult to accurately reflect the work performance of teachers and staff in the measurement of performance-related pay. Based on this background and trend, this paper conducted an in-depth discussion on the salary system in China’s colleges and universities, and used the support vector regression algorithm to study the performance-based salary management of colleges and universities based on linear programming and mathematical statistics. The research showed that with other conditions remaining unchanged, the proportions of primary, intermediate, deputy and senior high school students who were dissatisfied with the system of performance-based pay in colleges and universities were 82%, 57%, 46%, and 9%, respectively. After applying linear programming and mathematical statistics, the proportions of elementary, intermediate, sub-high and ortho-high became 1%, 2%, 5% and 7% respectively. The decline of the first three was particularly obvious, and the teachers with positive high school also dropped by 2%, indicating that linear programming and mathematical statistics were beneficial to the management of performance wages in colleges and universities.

A optimum prediction model of chloride ion diffusion coefficient of machine-made sand concrete based on different machine learning methods

The chloride ion diffusion coefficient (CIDC) is an important index to evaluate the durability performance of machined sand concrete. In this study, a database of CIDC is established for 82 sets of experimental data for 4 influence parameters. Meanwhile, different models with 4 input variables (Water to binder ratio, fly ash ratio, relative humidity and temperature) and 1 output variable (CIDC) were constructed to predict CIDC using extreme gradient boosting algorithm (XGBoost), backpropagation algorithm (BP), support vector regression algorithm (SVR), and bp algorithm optimized by Bayesian formula. Meanwhile, the SHapley Additive exPlanations (SHAP) method in python is introduced to characterize the established models to clarify the principles behind the model prediction. Based on the findings, it appears that the Bayesian-BP model is the one that performs the best in terms of making predictions. Furthermore, based on the prediction of the Bayesian-BP model, the percentage contribution of various variable factors in the model is evaluated by the SHAP analysis method, and the corresponding variables are characterized. It is found that the obtained characteristic variables with the CIDC have the same trend as the reality that verifies the reliability of the proposed optimum model. Moreover, CIDC was predicted by controlling variables, and flexible adjustments to CIDC were achieved based on this.

Predicting the final grade using a machine learning regression model: insights from fifty percent of total course grades in CS1 courses.

This article introduces a model for accurately predicting students' final grades in the CS1 course by utilizing their grades from the first half of the course. The methodology includes three phases: training, testing, and validation, employing four regression algorithms: AdaBoost, Random Forest, Support Vector Regression (SVR), and XGBoost. Notably, the SVR algorithm outperformed the others, achieving an impressive R-squared (R2) value ranging from 72% to 91%. The discussion section focuses on four crucial aspects: the selection of data features and the percentage of course grades used for training, the comparison between predicted and actual values to demonstrate reliability, and the model's performance compared to existing literature models, highlighting its effectiveness.

Solid-liquid phase transition temperature prediction of alloys based on machine learning key feature screening

A machine learning strategy is proposed based on the demand for prediction of solid-liquid phase transition temperature properties of multi-component precious metal alloys. Firstly, the candidate feature set are constructed by mathematical operation of the physical and chemical parameters of the material according to the chemical ratio of the alloy chemical formula. Then, a novel feature selection framework of correlation screening → genetic algorithm screening → feature weight ranking → exhaustive screening is proposed to effectively identify key feature combinations affecting solid-liquid phase transition temperature. Finally, the support vector regression algorithm is used to establish the "key feature combination-solid phase transition temperature" model with an error less than 9.83% and the "key feature combination-liquid phase transition temperature" model with an error less than 9.35%. The proposed new feature framework overcomes the inability of conventional feature selection techniques to simultaneously meet the requirements of interpretability, low computational complexity, and strong feature generality. Moreover, the comprehensive equilibrium complexity and accuracy (R2) constructed "solid temperature-key feature combination" with an error of less than 6.50% and "liquid phase transition temperature-key feature combination" with an error of less than 6.71 % by symbolic regression algorithm, which strengthened the understanding of the influence of key independent variable feature parameters on solid-liquid phase transition temperature prediction. This established machine learning strategy has excellent reliability in the prediction of solid-liquid phase transition temperature of multi-component alloys, and the proposed feature selection framework is expected to develop into a new feature selection technique.

A comparison of CAPM and Fama-French three-factor model under Machine Learning approaching

With the economy experiencing rapid growth in recent years, more individuals have started venturing into the stock market. Precisely forecasting the rate of return can mitigate investment risks for stock investors and significantly enhance their investment returns. The Capital Asset Pricing Model (CAPM) and the 3-factor Fama-French model (FF3) are widely recognized in academic and practical settings. This model comparison provides frameworks to analyze the relationship between portfolio risk and return in inefficient markets. This research utilized the Support Vector Regression (SVR) algorithm to forecast the returns of a diversified portfolio in the Hanoi stock market (HNX) from 2010 to 2022. Subsequently, the explanatory power of the CAPM and FF3 models were compared using the Ordinary Least Squares (OLS) algorithm. Finally, this research incorporated the SVR algorithm within the FF3 framework to develop a predictive model. The research findings demonstrate that the FF3 model provides a superior explanation to the CAPM model. Additionally, the study reveals that the SVR algorithm outperforms the OLS algorithm in terms of efficiency, as it yields lower Root Mean Square Error (RMSE) values. Consequently, the next research direction entails replacing the FF3 model with a more comprehensive multi-factor model, anticipating obtaining an enhanced predictive model.

P177: Motoric cognitive risk syndrome is associated with MRI-derived brain age: the Arakawa Geriatric Cohort Study

Objective:Motoric cognitive risk (MCR) syndrome is characterized by slow gait speed and subjective cognitive decline, which could predict future dementia. Previous research reported the associations between MCR and gray matter volume reduction in total and specific cortical regions and increased white matter hyperintensities in the brain. However, knowledge is scarce on the relationship between MCR and neuroimaging-derived brain age. The present study explored the association between MCR and brain-predicted age differences.Methods:The present study is a cross-sectional study that involved 1,099 community- dwelling older adults in the Arakawa Ward, Tokyo, Japan, who were between 65 to 84 years old in 2016. In defining MCR, subjective cognitive decline was deemed positive for those who answered “Yes” to the question, “Do you feel that you have more problems with memory than most?” in the Geriatric Depression Scale. Slower gait speed was defined as walking slower than the -1 standard deviation of the age- and sex- stratified gait speed. Brain age was predicted on the 1,021 participants with brain magnetic resonance imaging without severe artifacts or lesions. We used the support vector regression algorithm using MATLAB's “fitrsvm” function, applying ten-fold cross-validation to the results of primary component analysis of the spatially normalized gray-matter images. We calculated the brain-predicted age difference (Brain-PAD) by subtracting the chronological age from the predicted brain age. After excluding the participants with dementia (N=23), the difference in the mean Brain-PAD between MCR+ and MCR- was compared with the Student's t-test. The association between MCR and Brain PAD was examined with multiple regression analyses, adjusting for clinical-demographical data.Results:The median ages were 72 for both MCR+ (N=96) and MCR- (N=902). The mean Brain-PAD was 3.29 for MCR+ and -0.19 for MCR- (p < .001, Hedges' g = -0.504). Multiple regression analysis showed a significant association between Brain- PAD and MCR (standardized β = 0.159, p < .001) after adjusting for covariates.Conclusion:The present findings suggest that MCR reflects accelerated brain aging, which may increase the risk of neurodegeneration. Future studies should examine the longitudinal trajectories of brain age and incident dementia in participants with MCR.

Accelerated brain ageing in participants with higher cerebrovascular lesion load across aging, cognitive impairment, and mixed presentation

AbstractBackgroundWe aimed to investigate the association between cortical brain age and white matter hyperintensities (WMH) in diverse forms of clinically‐defined vascular conditions. Our hypothesis was that a higher WMH burden would be associated with higher cortical brain age in all clinical subtypes.MethodWe used standardized MRI data obtained from participants in the COMPASS‐ND cohort of the CCNA. We used a standard linear support vector regression algorithm to estimate brain age, matching chronological age to cortical anatomical measurements obtained with the FreeSurfer toolbox on T1‐weighted MRI, sex, and intracranial volume. We calculated brain‐PAD (i.e., predicted brain age minus real age) and applied bias adjustment to remove age‐dependency in the estimates. A validated automated technique utilizing T2‐weighted and fluid attenuated inversion recovery MRIs was used to compute WMH loads.ResultParticipants included 107 CIE, 240 mild cognitive impairment (MCI), 115 vascular‐MCI (V‐MCI), 81 probable Alzheimer’s disease (AD), and 50 V‐AD. There was a significant difference in brain‐PAD [ F (4,599) = 56, P < 0.001, ANCOVA test] among groups, whilst adjusting for sex and chronological age. All four categories of patients exhibited a significantly higher mean brain‐PAD than CIE (P < 0.001), with the AD cohort having the highest brain‐PAD. There was a significant difference in WMH loads [ F (4,586) = 48, P < 0.001, ANCOVA test] between groups, whilst adjusting for sex and age. All cohorts showed a positive correlation between brain‐PAD and WMH, indicating accelerated ageing in people with higher WMHs. We observed moderate and significant correlations between brain‐PAD and WMH loads for V‐MCI (r = 0.34, P < 0.001) and V‐AD (r = 0.36, P = 0.001) but weaker, while significant correlations in CIE (r = 0.20, P = 0.035) and AD cohorts (r = 0.24, P = 0.025).ConclusionWe observed a positive and significant link between brain‐PAD and WMH in all categories. This finding indicates on the importance of treatment and prevention strategies for vascular risk factors, which might be able to slow down the progression of cerebrovascular lesions and delay the effect on cortical thickness.

Cricket Players Selection for National Team and Franchise League using Machine Learning Algorithms

Cricket player selection is a crucial task for both national teams and franchise leagues. Traditionally, selectors rely on their experience and knowledge to evaluate a player's physical fitness, batting, and bowling performance. However, with the advancements in machine learning algorithms, it is possible to automate and improve the selection process. In this research, a Machine Learning (ML)-based approach is proposed for cricket player selection. This approach uses a combination of physical fitness data, batting and bowling statistics, and other relevant metrics to create a comprehensive player profile. Then three ML algorithms-linear regression, support vector regression, and random forests-are employed to identify the most promising players. Data is collected on a large number of cricket players and their performance in national and franchise leagues from the two most prominent cricket websites, espncricinfo, and cricbuzz, to evaluate the proposed approach. Then the proposed approach trained and tested the ML models on this data and compared their accuracy and performance. Based on the performance scores obtained from these models, two squads are selected for the national team and one squad for each franchise league team. The results demonstrate that the proposed approach can significantly improve the selection process and identify players with high potential. Furthermore, results found that the support vector regression algorithm outperformed other ML models in terms of prediction and player selection.

Single-point curved fiber optic pulse sensor for physiological signal prediction based on the genetic algorithm-support vector regression model

For hypertensive patients, a real-time human physiological signal monitoring system helps to track blood pressure status and provides valuable data to clinicians for early diagnosis and timely intervention. Optical fiber pulse sensing has superior conditions such as resistance to electromagnetic interference and richness of measured pulse characteristics. In this study, a monitoring system based on a single-point curved fiber pulse sensor (CFPS) was used to collect the pulse wave signal of a human radial artery. The features of the pulse wave signal were used to estimate the pulse wave transit time (PTT) and Blood Pressure—systolic blood pressure (SBP) and diastolic blood pressure (DBP)—based on support vector regression (SVR) optimized by a genetic algorithm (GA) (GA-SVR model). The results show that the root mean square error (RMSE) of SBP, DBP and PTT were 1.571 mmHg, 3.250 mmHg and 5.719 ms, respectively, and the results met the Advancement of Medical Instrumentation (AAMI) criteria. Therefore, a real-time pulse wave monitoring system based on a single-point CFPS can well predict human physiological signals such as PTT, SBP and DBP.

Greenhouse environmental control target constrained by discrete surface curvature and multi-objective optimization algorithm

Protected greenhouse provides a suitable micro-environment for crop growth, which enhances the quality and yield of agricultural products. How to optimize the greenhouse environment is currently a pressing issue in protected agriculture. In this research, we propose a new environmental control method from the perspectives of crop photosynthesis rate and environmental control cost. The photosynthesis rate data are collected from a nested experiment of tomatoes, and these data are combined with the support vector regression algorithm to establish a photosynthesis rate model. The cost function of greenhouse environmental control is established by theoretical calculation. The non-dominated solution sets of photosynthesis rate and control cost are obtained through an improved non-dominated sorting genetic algorithm Ⅱ. To determine the optimal solution among the numerous Pareto solutions, the geometric characteristics of the photosynthetic response surface are considered to constrain the control target. Then, a discrete surface curvature method is proposed to computer the constraint region. The intersection between the Pareto solutions and the constraint region is regarded as the environmental control target. The validation experiment demonstrates that the proposed method can effectively enhance photosynthetic accumulation and reduce control cost. Compared to the traditional saturation method, this method achieves a 53% cost reduction while still obtaining 89% photosynthetic accumulation. Compared to the U-chord method, this method increases photosynthetic accumulation by 12% with only a 3% increase in cost. The research has important implications for protected agriculture production.

An optimization approach for environmental control using quantum genetic algorithm and support vector regression

Photosynthesis serves as the foundation for vegetable crop yield. It is crucial to appropriately regulate the environmental parameters associated with photosynthesis to ensure efficient production and energy conservation in plant factories or greenhouses. In this research, we proposed a novel optimization approach for determining the target value of environmental control, aiming to balance plant growth and energy cost. By employing hydroponic lettuces as experimental samples, we measured their photosynthetic rates (Pn) under various combinations of four environmental factors: air temperature (AT), nutrient solution temperature (NST), photon flux density (PFD), and CO2 concentration ([CO2]). The photosynthetic data were combined with the support vector regression algorithm to develop a Pn prediction model. This model achieved a coefficient of determination of 0.9748, a root mean square error value of 0.9302 µmol∙m−2∙s−1, and a mean absolute error value of 1.1813 µmol∙m−2∙s−1. The model provide data for subsequent environmental control. The quantum genetic algorithm (QGA) was employed to search the optimal Pn and corresponding PFD, [CO2], and NST at different ATs. The fitness function for QGA was developed considering both the Pn and the energy consumption. This approach could calculate the target environments (PFD, [CO2], and NST) for any given AT. Compared with the Pn maximization approach, the energy cost-saving rate was 1.5 to 3.5 times higher than the Pn loss. The proposed approach could quickly and accurately determine an optimal environmental control target value, outperforming other approaches in complexity and generality. Thus, this study offers an elegant approach to environmental control for hydroponic cultivation.

Application of machine learning algorithms in drug screening

At present, in the medical field, drug screening is usually performed using in vivo drug experiments. However, it is very time-consuming and laborious to conduct in vivo experiments on a large number of drugs to be screened one by one. This paper attempts to propose using machine learning algorithms to perform preliminary screening of a large number of compounds to be screened and their molecular structures to reduce the workload of in vivo experiments. Among them, it is internationally recognized that there is an important association between breast cancer progression and the alpha subtype of the estrogen receptor. Anti-breast cancer drug candidates with excellent efficacy need to contain compounds that can better antagonize ERα activity. In this paper, the research object is narrowed down from compounds to the molecular structure of the compounds, and then the random forest regression algorithm is used to develop the molecular structure-ERα activity prediction model. Molecular structures with significant effects on biological activity were screened from molecular structure descriptors in numerous compounds. Four different kernel functions were used to conduct comparative experiments, and finally a support vector regression algorithm based on radial basis kernel function was established, which realized the quantitative prediction of compounds on biological activity of ERα, and could find potential compounds beneficial to breast cancer treatment. This is a novel, computer-based method for preliminary drug screening, which can help medical researchers effectively narrow the scope of experiments and achieve more accurate optimization of drugs.

Optimization studies on batch extraction of phenolic compounds from Azadirachta indica using genetic algorithm and machine learning techniques

Phenolic compounds play a crucial role as secondary metabolites due to their substantial biological activity and medicinal value. These compounds are present in various parts of plant species. This study focused on solid-liquid batch extraction to recover total phenolic compounds from Azadirachta indica leaves. The experimental design was based on the Taguchi L16 array, considering four independent factors: extraction time, temperature, particle size, and solid-to-solvent ratio. Among these factors, the particle size exerted the maximum influence. Particle size inversely affects the yield of total phenolic content (TPC), while temperature, time, and solid-to-liquid ratio have a direct impact. The process factors concerned were investigated both experimentally and through machine learning techniques. Support vector regression (SVR) and random forest method (RFM) algorithms were utilized for predicting TPC, while a genetic algorithm (GA) was employed to derive optimal process parameters. The GA predicts the optimal extraction factors, yielding the maximum TPC. During this study, these factors were the following: particle size of 0.15 mm, extraction time of 40 min, solid-to-liquid ratio of 1:25 g/mL, and a temperature of 55 °C, with a predicted value of 23.039 mg GAE/g of plant material. Notably, in this study, the SVR values of TPC yield closely matched the experimental values for the training and test data set when compared with the random forest method values.