ИСТИНА |
Войти в систему Регистрация |
|
ИСТИНА ИНХС РАН |
||
The analysis of precipitation, an important atmospheric parameter, in practice is complicated not only due to the nature of its variability, but also because of the presence of missing values that can be crucial for statistical and other techniques. We discuss handling the missing values in such data involving various machine learning algorithms to improve ``pure probabilistic'' pattern-based methodology. The examples of suggested methods implementations are demonstrated for the observations in Potsdam and Elista.