ABSTRACT
This chapter considers models that ignore the spatial information and are purely based on the time series data. It aims to describe wind speed, please bear in mind its direct applicability to wind power forecast. The forecasting can be performed either on wind speed or on wind power. Wind speeds are nonnegative and their distribution is right skewed. Support vector machine is one of the machine learning methods that are employed in wind speed forecasting. Support vector machine was initially developed for the purpose of classification, following and extending the work of optimal separating hyperplane. The data points that are more interior to a data class and farther away from the separating boundary do not affect the classification outcome. Support vector machine (SVM) for classification and SVM for regression use different loss functions. When training SVM and artificial neural network, the cross-validation strategy is used to decide the exogenous parameters.
