An Analysis of Turkey Wind Speed Data with the Marginal Entropy and a New Index about Prediction

Estimations and analyses of wind speed come into prominence since the wind energy is of great importance among renewable energy resources. From this point of view, Turkey wind speed data are analyzed and mapped via the marginal entropy. Marginal entropy or entropy phenomenon as stated in Information Theory is a measure of uncertainty of a probability distribution. In the study, the stations which they have the highest and lowest entropy values have been determined and data of the stations are used in order to predict wind speed values using the Support Vector Regression. The idea of the Support Vector Regression is based on the computation of a linear regression function in a high dimensional feature space. It attempts to minimize the generalization error bound so as to achieve generalized performance. Moreover, the Chaotic Approach has been implemented to prepare an input matrix for Support Vector Regression and the Singular Spectrum Analysis has been used to fill the gaps of the wind speed data. Performance criteria of the prediction and the entropy values are scrutinized so as to determine a relationship between the entropy and predictability. The Normalized Marginal Entropy has been proposed as an index that it represents the information carried from data and expressed as the ratio between a marginal entropy and maximum entropy. A high normalized marginal entropy means high information about data thanks to a surprise factor inside the data. Therefore the data with high normalized marginal entropy can be predicted more accurately. Results support the idea based on normalized marginal entropy, namely; the more entropy data have, the more predictable they are.