Search academic texts

Information × Registration Number 0219U005029, 0117U007281 , R & D reports Title Determination anomalies (outliers) in bivariate non-Gaussian data popup.stage_title Head Prikhodko Sergii Borisovich, Доктор технічних наук Registration Date 15-04-2019 Organization Kherson brench of the National University of Shipbuilding named after admiral Makarov popup.description2 Object of research - the process of determining anomalies (outliers) in bivariate non-Gaussian data. A research objective - increasing the reliability of determining outliers in bivariate non-Gaussian data by using of the bivariate normalizing Johnson transform, which will reduce the number of errors in determining outliers compared to existing methods. Research methods - methods of probability theory, the mathematical statistics, mathematical modeling. Scientific novelty of the research results: For the first time are obtained: 1. Equation of the transformed prediction ellipse based on the bivariate normalizing Johnson transformation of the SB family, which allows you to remove bivariate outliers in non-Gaussian data that have a bivariate Johnson distribution of the SB family, without applying the assumption of their normality and non-correlation. 2. Equation of the transformed prediction ellipse based on the bivariate normalizing Johnson transformation of the SU family, which allows you to remove bivariate outliers in non-Gaussian data that have bivariate Johnson distribution of the SU family, without applying the assumption of their normality and non-correlation. Product Description popup.authors К. С. Пугаченко Л. М. Макарова Н. В. Приходько О. О. Кудін Т. Г. Смикодуб popup.nrat_date 2020-04-02 Close