id: 02195566
dt: a
an: 02195566
au: Muhlenbach, Fabrice; Lallich, Stéphane; Zighed, Djamel A.
ti: Outlier handling in the neighbourhood-based learning of a continuous class.
so: Suzuki, Einoshin (ed.) et al., Discovery science. 7th international
    conference, DS 2004, Padova, Italy, October 2‒5, 2004. Proceedings.
    Berlin: Springer (ISBN 3-540-23357-1/pbk). Lecture Notes in Computer
    Science 3245. Lecture Notes in Artificial Intelligence, 314-321 (2004).
py: 2004
pu: Berlin: Springer
la: EN
cc: 
ut: 
ci: 
li: doi:10.1007/b100845
ab: Summary: This paper is concerned with the neighbourhood-based supervised
    learning of a continuous class. It deals with identifying and handling
    outliers. We first explain why and how to use the neighbourhood graph
    issued from predictors in the prediction of a continuous class. Global
    quality of the representation is evaluated by a neighbourhood
    autocorrelation coefficient derived from the Moran spatial
    autocorrelation coefficient. Extending the analogy with spatial
    analysis, we suggest to identify outliers by using the scattering
    diagram associated to the Moran coefficient. Several experiments
    realized on classical benchmarks show the interest of removing the
    outliers with this new method.
rv: