Approximate Sequential Patterns for Incomplete Databases

Abstract : Industrial databases often contains a large amount of unfilled information. During the knowledge discovery process one specific processing step is often necessary in order to remove these incomplete data either by deleting them or by assessing them. When the data mining task consists in mining frequent sequences, incomplete data are, most of the time, deleted, which leads to an important loss of information. Extracted knowledge becomes so less representative of the whole database. We thus propose a method that uses estimation of missing values contained into incomplete records, while computing the frequency of sequences.
Type de document :
Rapport
RR-07003, 2007
Liste complète des métadonnées

https://hal-lirmm.ccsd.cnrs.fr/lirmm-00129415
Contributeur : Celine Fiot <>
Soumis le : mercredi 7 février 2007 - 12:09:18
Dernière modification le : jeudi 11 janvier 2018 - 06:14:31

Identifiants

  • HAL Id : lirmm-00129415, version 1

Collections

Citation

Céline Fiot. Approximate Sequential Patterns for Incomplete Databases. RR-07003, 2007. 〈lirmm-00129415〉

Partager

Métriques

Consultations de la notice

38