Resumen

CASTRO AGUILAR, Gilberto Fernando et al. Outliers mining applications in project management organizations. Rev cuba cienc informat [online]. 2016, vol.10, suppl.1, pp.195-209. ISSN 2227-1899.

ABSTRACT Outliers mining are a data mining area of ​​data mining; have to do with detecting rare data or unusual behavior data. This discipline has high applicability in dissimilar scenarios among which include revenue assurance in telecommunications, financial fraud detection and security. In this paper the authors present different methods for the discovery outlier’s data on an approach that combines the techniques of outliers mining such as: supervised methods, unsupervised methods and semi-supervised ​​methods for outlier’s detection. The applicability of outliers mining is also presented in detecting errors and failures in the management of organizations oriented software development projects. In particular, these techniques could be used in revenue assurance process, in organizations oriented to software projects. Finally, authors presented results to apply, algorithm designed in "tasks and resources" data set published by Project Management Laboratory and generated GESPRO system. At the end, non-parametric statistical tests for comparing different algorithms, based on its detection performance. They arrive at conclusions and identify which algorithms presented the best performance.

Palabras clave : outliers mining; clusters; project management; software projects.

        · resumen en Español     · texto en Español     · Español ( pdf )