Search results
1 – 10 of over 25000
József Valyon and Gábor Horváth
The purpose of this paper is to present extended least squares support vector machines (LS‐SVM) where data selection methods are used to get sparse LS‐SVM solution, and to…
Abstract
Purpose
The purpose of this paper is to present extended least squares support vector machines (LS‐SVM) where data selection methods are used to get sparse LS‐SVM solution, and to overview and compare the most important data selection approaches.
Design/methodology/approach
The selection methods are compared based on their theoretical background and using extensive simulations.
Findings
The paper shows that partial reduction is an efficient way of getting a reduced complexity sparse LS‐SVM solution, while partial reduction exploits full knowledge contained in the whole training data set. It also shows that the reduction technique based on reduced row echelon form (RREF) of the kernel matrix is superior when compared to other data selection approaches.
Research limitations/implications
Data selection for getting a sparse LS‐SVM solution can be done in the different representations of the training data: in the input space, in the intermediate feature space, and in the kernel space. Selection in the kernel space can be obtained by finding an approximate basis of the kernel matrix.
Practical implications
The RREF‐based method is a data selection approach with a favorable property: there is a trade‐off tolerance parameter that can be used for balancing complexity and accuracy.
Originality/value
The paper gives contributions to the construction of high‐performance and moderate complexity LS‐SVMs.
Details
Keywords
Angela J. Black, David G. McMillan and Fiona J. McMillan
This paper aims to empirically test for multiple cointegrating vectors in a holistic manner. Theoretical developments imply bivariate cointegration among stock prices, dividends…
Abstract
Purpose
This paper aims to empirically test for multiple cointegrating vectors in a holistic manner. Theoretical developments imply bivariate cointegration among stock prices, dividends, output and consumption where independent models identify key theoretical cointegration vectors.
Design/methodology/approach
This paper considers both Johansen and Horvath–Watson testing approaches for cointegration. This paper also examines the forecasting power of these cointegrating relationships against alternate forecast variables.
Findings
The results suggest evidence of a long-run cointegrating relationship between stock prices, dividends, output and consumption, although not necessarily linked by a single common stochastic trend; each series responds to disequilibrium with greater evidence of a reaction from dividends and consumption – of note, output responds to changes in stock market equilibrium; and there is forecast power from the joint stock market–macro cointegrating vector for stocks returns and consumption growth over the historical average. Of particular note, other forecast models that include consumption perform well and suggest a key role for this variable in stock return and consumption growth forecasts.
Originality/value
This is the first paper to combine the cointegrating relationships between stocks, dividends, output and consumption. Thus, the empirical validity of stated theoretical hypotheses can be analysed. The forecast results also demonstrate the usefulness of this. They also show that forecast models that include consumption perform well and suggest a key role for this variable in stock return and consumption growth forecasts.
Details
Keywords
Thiago Turchetti Maia, Antônio Pádua Braga and André F. de Carvalho
To create new hybrid algorithms that combine boosting and support vector machines to outperform other known algorithms in selected contexts of binary classification problems.
Abstract
Purpose
To create new hybrid algorithms that combine boosting and support vector machines to outperform other known algorithms in selected contexts of binary classification problems.
Design/methodology/approach
Support vector machines (SVM) are known in the literature to be one of the most efficient learning models for tackling classification problems. Boosting algorithms rely on other classification algorithms to produce different weak hypotheses which are later combined into a single strong hypothesis. In this work the authors combine boosting with support vector machines, namely the AdaBoost.M1 and sequential minimal optimization (SMO) algorithms, to create new hybrid algorithms that outperform standard SVMs in selected contexts. This is achieved by integration with different degrees of coupling, where the four algorithms proposed range from simple black‐box integration to modifications and mergers between AdaBoost.M1 and SMO components.
Findings
The results show that the proposed algorithms exhibited better performance for most problems experimented. It is possible to identify trends of behavior bound to specific properties of the problems solved, where one may hence apply the proposed algorithms in situations where it is known to succeed.
Research limitations/implications
New strategies for combining boosting and SVMs may be further developed using the principles introduced in this paper, possibly resulting in other algorithms with yet superior performance.
Practical implications
The hybrid algorithms proposed in this paper may be used in classification problems with properties that they are known to handle well, thus possibly offering better results than other known algorithms in the literature.
Originality/value
This paper introduces the concept of merging boosting and SVM training algorithms to obtain hybrid solutions with better performance than standard SVMs.
Details
Keywords
Jin Gi Kim, Hyun-Tak Lee and Bong-Gyu Jang
This paper examines whether the successful bid rate of the OnBid public auction, published by Korea Asset Management Corporation, can identify and forecast the Korea…
Abstract
Purpose
This paper examines whether the successful bid rate of the OnBid public auction, published by Korea Asset Management Corporation, can identify and forecast the Korea business-cycle expansion and contraction regimes characterized by the OECD reference turning points. We use logistic regression and support vector machine in performing the OECD regime classification and predicting three-month-ahead regime. We find that the OnBid auction rate conveys important information for detecting the coincident and future regimes because this information might be closely related to deleveraging regarding default on debt obligations. This finding suggests that corporate managers and investors could use the auction information to gauge the regime position in their decision-making. This research has an academic significance that reveals the relationship between the auction market and the business-cycle regimes.
Details
Keywords
Harleen Kaur and Vinita Kumari
Diabetes is a major metabolic disorder which can affect entire body system adversely. Undiagnosed diabetes can increase the risk of cardiac stroke, diabetic nephropathy and other…
Abstract
Diabetes is a major metabolic disorder which can affect entire body system adversely. Undiagnosed diabetes can increase the risk of cardiac stroke, diabetic nephropathy and other disorders. All over the world millions of people are affected by this disease. Early detection of diabetes is very important to maintain a healthy life. This disease is a reason of global concern as the cases of diabetes are rising rapidly. Machine learning (ML) is a computational method for automatic learning from experience and improves the performance to make more accurate predictions. In the current research we have utilized machine learning technique in Pima Indian diabetes dataset to develop trends and detect patterns with risk factors using R data manipulation tool. To classify the patients into diabetic and non-diabetic we have developed and analyzed five different predictive models using R data manipulation tool. For this purpose we used supervised machine learning algorithms namely linear kernel support vector machine (SVM-linear), radial basis function (RBF) kernel support vector machine, k-nearest neighbour (k-NN), artificial neural network (ANN) and multifactor dimensionality reduction (MDR).
Details
Keywords
In order to improve the accuracy of project cost prediction, considering the limitations of existing models, the construction cost prediction model based on SVM (Standard Support…
Abstract
Purpose
In order to improve the accuracy of project cost prediction, considering the limitations of existing models, the construction cost prediction model based on SVM (Standard Support Vector Machine) and LSSVM (Least Squares Support Vector Machine) is put forward.
Design/methodology/approach
In the competitive growth and industries 4.0, the prediction in the cost plays a key role.
Findings
At the same time, the original data is dimensionality reduced. The processed data are imported into the SVM and LSSVM models for training and prediction respectively, and the prediction results are compared and analyzed and a more reasonable prediction model is selected.
Originality/value
The prediction result is further optimized by parameter optimization. The relative error of the prediction model is within 7%, and the prediction accuracy is high and the result is stable.
Details
Keywords
Yanjie Wang, Zhengchao Xie, InChio Lou, Wai Kin Ung and Kai Meng Mok
The purpose of this paper is to examine the applicability and capability of models based on a genetic algorithm and support vector machine (GA-SVM) and a genetic algorithm and…
Abstract
Purpose
The purpose of this paper is to examine the applicability and capability of models based on a genetic algorithm and support vector machine (GA-SVM) and a genetic algorithm and relevance vector machine (GA-RVM) for the prediction of phytoplankton abundances associated with algal blooms in a Macau freshwater reservoir, and compare their performances with an artificial neural network (ANN) model.
Design/methodology/approach
The hybrid models GA-SVM and GA-RVM were developed for the optimal control of parameters for predicting (based on the current month’s variables) and forecasting (based on the previous three months’ variables) phytoplankton dynamics in a Macau freshwater reservoir, MSR, which has experienced cyanobacterial blooms in recent years. There were 15 environmental parameters, including pH, SiO2, alkalinity, bicarbonate (HCO3−), dissolved oxygen (DO), total nitrogen (TN), UV254, turbidity, conductivity, nitrate (NO3−), orthophosphate (PO43−), total phosphorus (TP), suspended solids (SS) and total organic carbon (TOC) selected from the correlation analysis, with eight years (2001-2008) of data for training, and the most recent three years (2009-2011) for testing.
Findings
For both accuracy performance and generalized performance, the ANN, GA-SVM and GA-RVM had similar predictive powers of R2 of 0.73-0.75. However, whereas ANN and GA-RVM models showed very similar forecast performances, GA-SVM models had better forecast performances of R2 (0.862), RMSE (0.266) and MAE (0.0710) with the respective parameters of 0.987, 0.161 and 0.032 optimized using GA.
Originality/value
This is the first application of GA-SVM and GA-RVM models for predicting and forecasting algal bloom in freshwater reservoirs. GA-SVM was shown to be an effective new way for monitoring algal bloom problem in water resources.
Details
Keywords
Hualei Zhang and Mohammad Asif Ikbal
In response to these shortcomings, this paper proposes a dynamic obstacle detection and tracking method based on multi-feature fusion and a dynamic obstacle recognition method…
Abstract
Purpose
In response to these shortcomings, this paper proposes a dynamic obstacle detection and tracking method based on multi-feature fusion and a dynamic obstacle recognition method based on spatio-temporal feature vectors.
Design/methodology/approach
The existing dynamic obstacle detection and tracking methods based on geometric features have a high false detection rate. The recognition methods based on the geometric features and motion status of dynamic obstacles are greatly affected by distance and scanning angle, and cannot meet the requirements of real traffic scene applications.
Findings
First, based on the geometric features of dynamic obstacles, the obstacles are considered The echo pulse width feature is used to improve the accuracy of obstacle detection and tracking; second, the space-time feature vector is constructed based on the time dimension and space dimension information of the obstacle, and then the support vector machine method is used to realize the recognition of dynamic obstacles to improve the obstacle The accuracy of object recognition. Finally, the accuracy and effectiveness of the proposed method are verified by real vehicle tests.
Originality/value
The paper proposes a dynamic obstacle detection and tracking method based on multi-feature fusion and a dynamic obstacle recognition method based on spatio-temporal feature vectors. The accuracy and effectiveness of the proposed method are verified by real vehicle tests.
Details
Keywords
Yanmei Huang, Changrui Deng, Xiaoyuan Zhang and Yukun Bao
Despite the widespread use of univariate empirical mode decomposition (EMD) in financial market forecasting, the application of multivariate empirical mode decomposition (MEMD…
Abstract
Purpose
Despite the widespread use of univariate empirical mode decomposition (EMD) in financial market forecasting, the application of multivariate empirical mode decomposition (MEMD) has not been fully investigated. The purpose of this study is to forecast the stock price index more accurately, relying on the capability of MEMD in modeling the dependency between relevant variables.
Design/methodology/approach
Quantitative and comprehensive assessments were carried out to compare the performance of some selected models. Data for the assessments were collected from three major stock exchanges, namely, the standard and poor 500 index from the USA, the Hang Seng index from Hong Kong and the Shanghai Stock Exchange composite index from China. MEMD-based support vector regression (SVR) was used as the modeling framework, where MEMD was first introduced to simultaneously decompose the relevant covariates, including the opening price, the highest price, the lowest price, the closing price and the trading volume of a stock price index. Then, SVR was used to set up forecasting models for each component decomposed and another SVR model was used to generate the final forecast based on the forecasts of each component. This paper named this the MEMD-SVR-SVR model.
Findings
The results show that the MEMD-based modeling framework outperforms other selected competing models. As per the models using MEMD, the MEMD-SVR-SVR model excels in terms of prediction accuracy across the various data sets.
Originality/value
This research extends the literature of EMD-based univariate models by considering the scenario of multiple variables for improving forecasting accuracy and simplifying computability, which contributes to the analytics pool for the financial analysis community.
Details