Traditional Support Vector Regression Research Articles

The conditional probability of travel speed and its application to short-term prediction

Over the past a few decades, a large number of researchers have worked on embedding mathematical methods into the prediction of traffic conditions. However, a lot of previous work only considered the temporal and spatial covariates as static explanatory variables. Different from previous studies, this paper investigates the conditional probability of freeway travel speed based on its lag values. Such conditional probability can be regarded as an attribute for each road segment. Based on this attribute, this paper proposes a conditional cumulative distribution function (CCDF)-based travel speed prediction approach. A real-world case study is conducted to predict short-term freeway speeds. The proposed CCDF-based model is compared with the Neural Networks (NN) model and the Support Vector Regression (SVR) model. The proposed model is faster in model training. Moreover, it can significantly improve the prediction accuracy, with 7.11% accuracy increase, compared to traditional NN and SVR prediction approaches.

Application of Data Mining Technology Based on Wireless Sensor Networks in Oceanographic Forecasting

This paper aims to present a desirable prediction method for oceanographic trends. Therefore, an online monitoring scheme was prepared to collect the accurate oceanographic hydrological data based on wireless sensor network (WSN) and computer technology. Then, the data collected by the WSN were processed by support vector regression algorithm. To obtain the most important parameters of the algorithm, the particle swarm optimization was introduced to search for the global optimal solution through the coopetition between the particles. After that, an oceanographic hydrological data collection and observation system was created based on the hydrological situation of New York harbour. Then, the traditional support vector regression and the proposed method were applied to predict the oceanographic trends based on water temperature, salinity and other indices. The results show that the proposed algorithm enhanced the data utilization rate of the WSN, and achieved good prediction accuracy. The research provides important insights into the application of advanced technology in oceanographic forecast.

Forecasting government bond spreads with heuristic models: evidence from the Eurozone periphery

This study investigates the predictability of European long-term government bond spreads through the application of heuristic and metaheuristic support vector regression (SVR) hybrid structures. Genetic, krill herd and sine–cosine algorithms are applied to the parameterization process of the SVR and locally weighted SVR (LSVR) methods. The inputs of the SVR models are selected from a large pool of linear and non-linear individual predictors. The statistical performance of the main models is evaluated against a random walk, an Autoregressive Moving Average, the best individual prediction model and the traditional SVR and LSVR structures. All models are applied to forecast daily and weekly government bond spreads of Greece, Ireland, Italy, Portugal and Spain over the sample period 2000–2017. The results show that the sine–cosine LSVR is outperforming its counterparts in terms of statistical accuracy, while metaheuristic approaches seem to benefit the parameterization process more than the heuristic ones.

A Mixed-Kernel-Based Support Vector Regression Model for Automotive Body Design Optimization Under Uncertainty

Finite element (FE) models are commonly used for automotive body design. However, even with increasing speed of computers, the FE-based simulation models are still too time-consuming when the models are complex. To improve the computational efficiency, support vector regression (SVR) model, a potential approximate model, has been widely used as the surrogate of FE model for crashworthiness optimization design. Generally, in the traditional SVR, when dealing with nonlinear data, the single kernel function-based projection cannot fully cover data distribution characteristics. In order to eliminate the application limitations of single kernel SVR, a method for reliability-based design optimization (RBDO) based on mixed-kernel-based SVR (MKSVR) is proposed in this research. The mixed kernel is constructed based on the linear combination of radial basis kernel function and polynomial kernel function. Through the particle swarm optimization (PSO) algorithm, the parameters of the mixed kernel SVR are optimized. The proposed method is demonstrated through a representative analytical RBDO problem and a vehicle lightweight design problem. And the comparitive studies for SVR and MKSVR in application indicate that MKSVR surpasses SVR in model accuracy.

Cloud Computing and Extreme Learning Machine for a Distributed Energy Consumption Forecasting in Equipment-Manufacturing Enterprises

Abstract Energy consumption forecasting is a kind of fundamental work of the energy management in equipment-manufacturing enterprises, and an important way to reduce energy consumption. Therefore, this paper proposes an intellectualized, short-term distributed energy consumption forecasting model for equipment-manufacturing enterprises based on cloud computing and extreme learning machine considering the practical enterprise situation of massive and high-dimension data. The analysis of the real energy consumption data provided by LB Enterprise was undertaken and corresponding calculating experiments were completed using a 32-node cloud computing cluster. The experimental results show that the energy consumption forecasting accuracy of the proposed model is higher than the traditional support vector regression and the generalized neural network algorithm. Furthermore, the proposed forecasting algorithm possesses excellent parallel performance, overcomes the shortcoming of a single computer’s insufficient computing power when facing massive and high-dimensional data without increasing the cost.

European Exchange Trading Funds Trading with Locally Weighted Support Vector Regression

In this paper, two different Locally Weighted Support Vector Regression (wSVR) algorithms are generated and applied to the task of forecasting and trading five European Exchange Traded Funds. The trading application covers the recent European Monetary Union debt crisis. The performance of the proposed models is benchmarked against traditional Support Vector Regression (SVR) models. The Radial Basis Function, the Wavelet and the Mahalanobis kernel are explored and tested as SVR kernels. Finally, a novel statistical SVR input selection procedure is introduced based on a principal component analysis and the Hansen, Lunde, and Nason (2011) model confidence test. The results demonstrate the superiority of the wSVR models over the traditional SVRs and of the v-SVR over the ε-SVR algorithms. We note that the performance of all models varies and considerably deteriorates in the peak of the debt crisis. In terms of the kernels, our results do not confirm the belief that the Radial Basis Function is the optimum choice for financial series.

Illumination correction of dyeing products based on Grey-Edge and kernel extreme learning machine

Changes in illumination will result in serious color difference evaluation error in the process of textile printing. In order to solve the problem, a novel illuminant estimation method based on kernel extreme learning machine (KELM) is proposed. Furthermore, a new efficient and low dimensional color feature extraction method based on Grey-Edge framework is adopted to replace the traditional high dimensional binary chromaticity histogram, which is used to represent the input data of KELM. The experiments show that the proposed color constancy method performs better than the traditional support vector regression (SVR) and basic extreme learning machine (ELM) based color constancy methods. Compared with SVR and ELM, the proposed method reduces the median and root mean square errors with approximately 6%, 11%, 43% and 48%, respectively.

PSO-SVR-Based Resource Demand Prediction in Cloud Computing

The essential of cloud computing is to offer elastic resources (such as CPU, memory, storage, and more) allocation to cloud customers on demand, and the resources are allocated dynamically in a pay-as-you-go fashion. In order to achieve this goal automatically while guaranteeing the performance of the application deployed in the cloud, a proactive resource scaling strategy is necessary for cloud providers. In this paper, we present an optimal resource usage prediction approach based on Support Vector Regression (SVR) that predicts resource demands from users in the near future. In order to improve the forecasting accuracy, Particle Swarm Optimization (PSO) is integrated in the model selection process for SVR to optimize the parameters of the model. Experiment results show that the prediction model achieves high accuracy and outperforms traditional SVR and Linear Regression (LR).

An adaptive support vector regression method for structural system reliability assessment and its application to a cable-stayed bridge

Engineering structures are most statically indeterminate structures consisting of various types of components and their failure modes exhibit randomness under random loads. A new adaptive support vector regression method is proposed for structural system reliability assessment. Compared with traditional support vector regression, the proposed adaptive support vector regression method involves two updating procedures to estimate structural limit state functions. Three verification examples involving a nonlinear limit state function, a truss bridge structure, and a geometrically nonlinear suspended structure are provided to illustrate the accuracy and the efficiency of the adaptive support vector regression method. A pre-stressed concrete cable-stayed bridge is utilized to demonstrate the applicability of the proposed method. The verification studies show that the proposed adaptive support vector regression method is an efficient method with reasonable accuracy for problems where closed-form failure functions are not available and the failure sequences exist in the structural system. The main failure sequences of the cable-stayed bridge are identified. The application studies of the cable-stayed bridge indicate that (a) the foremost failure sequence is the strength failure of side-span cables followed by the bending failure of towers and (b) the secondary failure sequence is the strength failure of mid-span cables followed by bending failure of mid-span girders.

A Hybrid Model of EMD and PSO-SVR for Short-Term Load Forecasting in Residential Quarters

Short-term load forecasting plays a vital role in the daily operational management of power utility. To improve the forecasting accuracy, this paper proposes a hybrid EMD-PSO-SVR forecasting model for short-term load forecasting based on empirical mode decomposition (EMD), support vector regression (SVR), and particle swarm optimization (PSO), also considering the effects of temperature, weekends, and holidays. EMD is used to decompose the residential electric load data into a number of intrinsic mode function (IMF) components and one residue; then SVR is constructed to forecast these IMFs and residual value individually. In order to gain optimization parameters of SVR, PSO is implemented to automatically perform the parameter selection in SVR modeling. Then all of these forecasting values are reconstructed to produce the final forecasting result for residential electric load data. Compared with the results from the EMD-SVR model, traditional SVR model, and PSO-SVR model, the result indicates that the proposed EMD-PSO-SVR model performs more effectively and more stably in forecasting the residential short-term load.

Integrating Support Vector Regression and a geomorphologic Artificial Neural Network for daily rainfall-runoff modeling

In spite of the efficiency of the Artificial Neural Networks (ANNs) for modeling nonlinear and complicated rainfall-runoff (R-R) process, they suffer from some drawbacks. Support Vector Regression (SVR) model has appeared to be a powerful alternative to reduce some of these drawbacks while retaining many strengths of ANNs. In this paper, to form a new rainfall-runoff model called SVR-GANN, a SVR model is combined with a geomorphologic-based ANN model. The GANN is a three-layer perceptron model, in which the number of hidden neurons is equal to the number of possible flow paths within a watershed and the connection weights between hidden layer and output layer are specified by flow path probabilities which are not updated during the training process. The capabilities of the proposed SVR-GANN model in simulating the daily runoff is investigated in a case study of three sub-basins located in a semi-arid region in Iran. The results of the proposed model are compared with those of ANN-based back propagation algorithm (ANN-BP), traditional SVR, ANN-based genetic algorithm (ANN-GA), adaptive neuro-fuzzy inference system (ANFIS), and GANN from the standpoints of parsimony, equifinality, robustness, reliability, computational time, simulation of hydrograph ordinates (peak flow, time to peak, and runoff volume) and also saving the main statistics of the observed data. The results show that prediction accuracy of the SVR-GANN model is usually better than those of ANN-based models and the proposed model can be applied as a promising, reliable, and robust prediction tool for rainfall-runoff modeling.

The Trend of Average Unit Price in Taipei City

The volatility of real estate prices is one of the key factors to the decision making of financial institutions, as well as to a country’s economic development. Therefore how to catch the trend of this real estate price has been an important issue for Governments and financial institutions. In this study, we discuss the trend of average unit price in a capital city, in hope of establishing a decent predicting model and key factors for this price. Other than traditional statistic methods, Neural Networks (NN) and Support Vector Regression (SVR) have demonstrated their advantages in previous research, and thus are applied and compared in this study. Variables are first summarized and concluded from earlier research and then selected by stepwise procedure. The result shows that SVR outperformed NN and stepwise procedure is valid in variable selections, and the key factors are previous trading price, Money supply M2 and New House-purchasing Loans.

An Adaptive Support Vector Regression Machine for the State Prognosis of Mechanical Systems

Due to the unsteady state evolution of mechanical systems, the time series of state indicators exhibits volatile behavior and staged characteristics. To model hidden trends and predict deterioration failure utilizing volatile state indicators, an adaptive support vector regression (ASVR) machine is proposed. In ASVR, the width of an error-insensitive tube, which is a constant in the traditional support vector regression, is set as a variable determined by the transient distribution boundary of local regions in the training time series. Thus, the localized regions are obtained using a sliding time window, and their boundaries are defined by a robust measure known as the truncated range. Utilizing an adaptive error-insensitive tube, a stabilized tolerance level for noise is achieved, whether the time series occurs in low-volatility regions or in high-volatility regions. The proposed method is evaluated by vibrational data measured on descaling pumps. The results show that ASVR is capable of capturing the local trends of the volatile time series of state indicators and is superior to the standard support vector regression for state prediction.

Short-Term Wind Speed Forecasting Study and Its Application Using a Hybrid Model Optimized by Cuckoo Search

The support vector regression (SVR) and neural network (NN) are both new tools from the artificial intelligence field, which have been successfully exploited to solve various problems especially for time series forecasting. However, traditional SVR and NN cannot accurately describe intricate time series with the characteristics of high volatility, nonstationarity, and nonlinearity, such as wind speed and electricity price time series. This study proposes an ensemble approach on the basis of 5-3 Hanning filter (5-3H) and wavelet denoising (WD) techniques, in conjunction with artificial intelligence optimization based SVR and NN model. So as to confirm the validity of the proposed model, two applicative case studies are conducted in terms of wind speed series from Gansu Province in China and electricity price from New South Wales in Australia. The computational results reveal that cuckoo search (CS) outperforms both PSO and GA with respect to convergence and global searching capacity, and the proposed CS-based hybrid model is effective and feasible in generating more reliable and skillful forecasts.

Modeling Method Based on Smooth Support Vector Regression with Independent Component Analysis Feature Extraction

Smooth Support Vector Regression (SSVR) is new modified edition of traditional support vector regression for better performance. To further improve the modeling capability of SSVR, it is necessary to take into account the feature extraction based on Independent Component Analysis (ICA) before SSVR. Simulation on the example of function approximation shows that the result of SSVR based on ICA feature extraction is better than that of SSVR without ICA preprocess.

Research of least squares support vector regression based on differential evolution algorithm in short-term load forecasting model

To improve the accuracy of short-term load forecasting, a differential evolution algorithm (DE) based least squares support vector regression (LSSVR) method is proposed in this paper. Through optimizing the regularization parameter and kernel parameter of the LSSVR by DE, a short-term load forecasting model which can take load affected factors such as meteorology, weather, and date types into account is built. The proposed LSSVR method is proved by implementing short-term load forecasting on the real historical data of Yangquan power system in China. The average forecasting error is less than 1.6%, which shows better accuracy and stability than the traditional LSSVR and Support vector regression. The result of implementation of short-term load forecasting demonstrates that the hybrid model can be used in the short-term forecasting of the power system more efficiently.

Online estimation of workpiece height by least squares support vector machines in wire electrical discharge machining

When machining workpieces with complicated and intricate shapes by wire electrical discharge machining, the workpiece height usually varies along a machining path. To ensure a stable and efficient machining, the machining parameters must be appropriately tuned based on the estimated workpiece height. Though an offline workpiece height estimation model can be established by traditional support vector regression with a satisfactory accuracy, the offline model is unable to cover all possible machining conditions. In this paper, least squares support vector machine is proposed to build up online a workpiece height estimation model. Due to the use of equality constraints in the formulation of least squares support vector machine, all data points are treated as support vectors, thus sparsity is lost as compared with traditional support vector machine. For online applications, a low computational load as well as a limited memory storage are required for an online algorithm. To meet the requirements of online workpiece height estimation, two measures are thus adopted. One is the use of a projection method which measures the relevance of a new data point with existing basic vectors by calculating its residue. The residue is used as criteria for admission of the new data point as a new member of the basic vector set. The other is the restriction on the size of the basic vector set. Removal of an insignificant basic vector is determined by its contribution to the model, which is measured by its coefficient in the model (or support value). Experimental results show that, by online learning, the estimation model can achieve an estimation error less than 2 mm at smooth parts of a workpiece. Based on the estimated workpiece height, proper machining parameters can be set and a stable machining can be achieved.

The combination of self-organizing feature maps and support vector regression for solving the inverse ECG problem

Noninvasive electrical imaging of the heart aims to quantitatively reconstruct transmembrane potentials (TMPs) from body surface potentials (BSPs), which is a typical inverse problem. Classically, electrocardiography (ECG) inverse problem is solved by regularization techniques. In this study, it is treated as a regression problem with multi-inputs (BSPs) and multi-outputs (TMPs). Then the resultant regression problem is solved by a hybrid method, which combines the support vector regression (SVR) method with self-organizing feature map (SOFM) techniques. The hybrid SOFM–SVR method conducts a two-step process: SOFM algorithm is used to cluster the training samples and the individual SVR method is employed to construct the regression model. For each testing sample, the cluster operation can effectively improve the efficiency of the regression algorithm, and also helps the setup of the corresponding SVR model for the TMPs reconstruction. The performance of the developed SOFM–SVR model is tested using our previously developed realistic heart-torso model. The experiment results show that, compared with traditional single SVR method in solving the inverse ECG problem, the proposed method can reduce the cost of training time and improve the reconstruction accuracy in solving the inverse ECG problem.

A hybrid economic indices based short-term load forecasting system

The bankruptcy event of Lehman Brothers and the corresponding global economic recession in 2008 and 2009, influenced the electricity load demand patterns for which traditional load forecasting approaches were not able to effectively predict. To overcome this problem, this paper proposes a new hybrid economic indices based short-term load forecasting (HEI-STLF) system. In which business indicators, such as the leading index or the coincide index, each combined with stock index as hybrid economic indices influencing factors for the support vector regression (SVR) model, to respond to the economic dynamics and reduce its impact on forecasting accuracy. The Taiwan island-wide electricity load demands from 2008 to 2011 are used as the case study for performance testing with different combinations of the Taiwan business indicator and the Taiwan Stock Exchange Capitalization-Weighted Stock Index (TAIEX). The results show that the proposed HEI-STLF system with hybrid economic indices of an annualized six-month rate of change of composite leading index and a 90days moving average of TAIEX, achieves the best forecasting performance. Compared to the traditional SVR load forecasting approach, it improves the forecasting accuracy in the best condition by 30.39% in the period when the load demands are affected by the global economic recession.

Optimization of molecular docking scores with support vector rank regression

This work introduces the support vector rank regression (SVRR) algorithm for the optimization of molecular docking scores. Seven original docking scores reported by two docking software were integrated by the SVRR algorithm. The resulting SVRR scores showed an average of 12.1% improvement (59.5-66.7%) in binding conformation prediction tests to rank the correctly computed conformation in the first place, along with 16.7% RMSD improvement (2.5414 vs. 2.1162 Å) for the top ranked conformations. In compound library screening (LS) tests, an average of 46.3% improvement (18.2-26.6%) was also observed to rank the correct ligand in the first place. Furthermore, it was shown that SVRR scores trained with different example datasets, using different training strategies, all exhibited exceedingly consistent accuracies, suggesting that the SVRR algorithm is highly robust and generalizable. In contrast, using the same training datasets, traditional support vector classification and regression algorithms failed to improve comparably the accuracy of LS and conformation prediction. These results suggested that, with additional features to indicate the comparative fitness between computed binding conformations, the SVRR algorithm holds the potential to create a new category of more accurate integrative docking scores.