A METHOD OF FORMALIZATION OF THE FEATURE SPACE FOR CLASSIFICATION OF THE FORECAST WEATHER CONDITION BY LOCAL FACTORS
Keywords:
feature space, local features, formalization, knowledge base, forecast, classification, dichotomization, weather change, navigation safetyAbstract
The aim of the article is to develop a method of formalization of the feature space for classification of the forecast weather condition by local factors in certain territories (areas). This goal is achieved by analysing the sources of information on approaches to the formalization of local weather features in different areas for their further selection and classification. The principle of determinism is proposed and implemented to formalize the description of local features of the forecast weather condition. In the formation of the feature space for the classification of the forecast weather condition according to local factors in certain areas (sites), group features of the dynamics of weather change are determined; their dichotomization is conducted and a reference set of features for each class of weather type is determined. The implementation of the matrix of correspondence of features and indices of belonging to weather types in the binary feature space allowed to move from linguistic variables in the knowledge base to Boolean ones. The most significant result is the development of a method of formalization of the feature space for the classification of the forecast weather condition according to local factors in certain territories (areas), which is distinguished by the use of binary relations and divisive procedures; hierarchical structure of the classification of weather types; features of the structure of the singleton knowledge base of local weather signs and its changes. The further direction of work is the formalization of local weather signs of another type with more complex properties; analysis of the completeness and reliability of input information to determine the advantage of the proposed solution options; implementation of a logical inference mechanism that will correspond to the chosen formalization apparatus, with the formation of a vector of "trust" in the solution options.
