Виявлення та оброблення невизначеностей у формі неповних даних методами інтелектуального анализу

• Дата: 2016
• Автор: Kuznietsova, Nataliia V.
• Формат: Стаття
• Мова: Ukrainian
• Опубліковано: The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" 2016
• Теми: risks analysis; data mining; Bayesian networks; informational technologies; system analysis; анализ рисков; методы анализа данных; сети Баеса; информационные технологии; системный анализ; аналіз ризиків; методи аналізу даних; мережі Байєса; інформаційні технології; системний аналіз
• Онлайн доступ: http://journal.iasa.kpi.ua/article/view/75213
• Назва журналу: System research and information technologies

In this paper, the methods for processing missing data are reviewed. The classification of methods depending on input data, data types and formats, and causes of data incompleteness associated with influence of uncertainties of the outside world and modeling object is proposed. The commonalities and differences between existing methods are investigated. The application peculiarities of these methods for filling missing data depending on properties of uncertainties are determined. It is shown that the traditional approach for filling the missing data by average values does not allow obtaining correct forecasts in many cases due to changes in sample’s properties. The usage of data mining methods technologies for dealing with missing data is proposed. An example of using regression methods is shown for filling missing data, in particular, using the forecast evaluation.