Rafael Castaneda,  Cláudia Ferlin,  Ronaldo Goldschmidt,  Jorge de Abreu Soares,  Luís Alfredo V. de Carvalho,  Ricardo Choren. 

Knowledge discovery in databases usually face the problem of missing values. Thus there are several preprocessing mechanisms that aim to make data imputation. However, these mechanisms normally deal with univariate cases, i.e. cases that present missing values in only one column. Iterative imputation mechanisms are capable of dealing with cases that present missing values in several columns, imputing values for one column at a time, but offer several implementation possibilities, from which the data analists find it difficult to choose. This paper presents a workflow-based platform to allow the easy setup, experimentation, and analisys of several iterative imputation techniques. It shows the usage of the platform and a sample experiment.

http://www.lbd.dcc.ufmg.br:8080/colecoes/sbbd/2008/017.pdf

Processando consulta.