Substantial work has been conducted on data mining in data warehouses as reported
in Han and Kamber (2000). This includes (but is not limited to) exception
detection in dimensional datasets (Knorr, Ng, & Tucakov, 2000), cubegrade generation
(Imielinski et al., 2002), constrained gradient analysis (Dong et al., 2001),
and discovery-driven examination of cubes (Sarawagi et al., 1998). Cubegrades are
association rules which express the impact of cube changes on a set of measures.
In the context of the governance running example, cubegrades can help find (1)
how the average asset held by enterprises is affected by the presence of females
on the board, or (2) how the enterprises having a good governance index (70%
and more) compare with enterprises with a lower governance index in terms of the
average amount of assets. In Naouali and Missaoui (2005), an approach towards
approximating the answer to OLAP queries and the identification of classification
and characteristic rules is proposed using the rough set theory.
Data.Mining.Techniques. for............
Data.Warehousing
In this section we focus on association rule mining from data cubes.
Rule Mining from Cubes
Association rule mining (ARM) in data cubes is different from the classical ARM
because measures in data cubes are aggregated values which depend intimately on
the value taken by each one of the cube dimensions, while each item (respectively
attribute) in transaction databases (resp.
Pages:
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489