Search academic texts

Information × Registration Number 0215U003591, 0113U006183 , R & D reports Title Methods and algorithms for recovery and enrichment of data in tasks of modeling online communities popup.stage_title Head Sydorenko Valeriy, Registration Date 26-02-2015 Organization Kremenchuk Mykhaylo Ostrogradskiy State University popup.description2 A brief analysis of urgency and current state of quality problems with data from social networks and their suitability for analysis was made. Typical features of the data, that complicate and may affect the reliability of modelling results, were shown. The principal methods of uninformative indicators enrichment and missing values restoration (imputation) were examined. The choice of enrichment and restoration methods was grounded on the basis of specific nature of social networking accounts. A flowchart of ensembles of models for data improvement on basis of algorithms for missing values imputation and uninformative factors enrichment was developed with use of factor analysis, clustering, association rules, trees and random forests. This approach is based on the paradigm of a combination of initial clustering models followed the imputation models and can form the basis for an information technology of improving the quality of initial data from social networks, what can help to increase the efficiency of Data Mining-projects in the tasks of communities modelling in social networks. Product Description popup.authors Слабченко Олеся Олегівна popup.nrat_date 2020-04-02 Close