Preview of Java Data Mining 2.0
Without expectations, there's no future, only an endless present.
—François Jacob, winner of the Nobel Prize in medicine
As the first version of the Java Data Mining (JDM) standard (JSR-73)
was being released, a new JSR, JSR-247, was submitted to continue
the work started in JSR-73. Although the features for JDM 2.0 are
still in flux, we have many expectations for its capabilities and use.
At the time of this writing, JDM 2.0 is actively being specified, and
the early draft review [JDM 2.0 2006], as specified through the Java
Community Process (JCP), has been completed.
As with most software projects and specifications, some features
do not make the cut for the first release. JSR-73 was no different. The
JDM expert group identified many features for inclusion in the first
release, but time and resource limitations required paring down the
feature set to a manageable size that still reflected a range of com-
monly used data mining techniques. For readers interested in more
detail of the JDM 2.0 specification, the latest available version can be
downloaded at www.jcp.org.
With JDM 2.0, the goals are, in part, to expand the set of mining
functions and algorithms to enable solving a broader range of data
mining problems. These include time series analysis using ARIMA
modeling, feature extraction using non-negative matrix factorization