Wind speed or wind power forecasting plays an important role in large-scale wind power penetration due to their uncertainty. Support vector regression, widely used in wind speed or wind power forecasting, aims at discovering natural structures of wind variation hidden in historical data. Most current regression algorithms, including least squares support vector regression (SVR), assume that the noise of the data is Gaussian with zero mean and the same variance. However, it is discovered that the uncertainty of short-term wind speed satisfies Gaussian distribution with zero mean and heteroscedasticity in this work. This kind of task is called heteroscedastic regression. In order to deal with this problem, we derive an optimal loss function for heteroscedastic regression and develop a new framework of v-SVR for learning tasks of Gaussian noise (GN) with heteroscedasticity. In addition, we introduce the stochastic gradient descent (SGD) method to solve the proposed model, which leads the models to be trained online. Finally, we reveal the uncertainty properties of wind speed with two real-world datasets and test the proposed algorithms on these data. The experimental results confirm the effectiveness of the proposed model.
Clustering is one of the most important topics in data mining and machine learning. The density peaks clustering (DPC) algorithm is a well-known density-based clustering method that can efficiently and effectively deal with non-spherical clusters. However, the computational methods of the local density and the distance measure are simple and easily ignore the correlation and the similarity between samples, and the manual setting of parameters has a great influence on the clustering results; therefore, the clustering performance of DPC is poor on the high-dimensional datasets. To address these issues, this paper presents an adaptive DPC algorithm with Fisher linear discriminant for the clustering of complex datasets, called ADPC-FLD. First, the kernel density estimation function is introduced to calculate the local density of the sample points. Pearson correlation coefficient between samples as a weight is employed to construct a weighted Euclidean distance function to measure the distance between samples. This considers both the spatial structure and the correlation of the samples. Then, a novel density estimation entropy is proposed, and based on the minimization of density estimation entropy, the density estimation parameters are adaptively selected according to the distribution characteristics of the data, which can efficiently eliminate the influence of manual setting. Third, an adaptive strategy of cluster center selection is designed to avoid the error caused by the noise data as the cluster centers and the uncertainty of manually selecting the cluster centers. Finally, Fisher linear discriminant algorithm is used to eliminate the irrelevant information and reduce the dimensionality of high-dimensional data, following on which an adaptive DPC method is implemented on six synthetic datasets, thirteen UCI datasets and seven gene expression datasets for comparing with other related algorithms. Experimental results on twenty-six datasets show that the proposed algorithm significantly outperforms several outstanding clustering approaches in terms of clustering accuracy and efficiency.
Attribute reduction as an important preprocessing step for data mining, and has become a hot research topic in rough set theory. Neighborhood rough set theory can overcome the shortcoming that classical rough set theory may lose some useful information in the process of discretization for continuous-valued data sets. In this paper, to improve the classification performance of complex data, a novel attribute reduction method using neighborhood entropy measures, combining algebra view with information view, in neighborhood rough sets is proposed, which has the ability of dealing with continuous data whilst maintaining the classification information of original attributes. First, to efficiently analyze the uncertainty of knowledge in neighborhood rough sets, by combining neighborhood approximate precision with neighborhood entropy, a new average neighborhood entropy, based on the strong complementarity between the algebra definition of attribute significance and the definition of information view, is presented. Then, a concept of decision neighborhood entropy is investigated for handling the uncertainty and noisiness of neighborhood decision systems, which integrates the credibility degree with the coverage degree of neighborhood decision systems to fully reflect the decision ability of attributes. Moreover, some of their properties are derived and the relationships among these measures are established, which helps to understand the essence of knowledge content and the uncertainty of neighborhood decision systems. Finally, a heuristic attribute reduction algorithm is proposed to improve the classification performance of complex data sets. The experimental results under an instance and several public data sets demonstrate that the proposed method is very effective for selecting the most relevant attributes with great classification performance.