Any body can explain me the unsupervised and
I try to discretize muneric attributes to use them in Classification (to
construct a decision tree)
in fact, when I use in Filter -> supervised --> disrcetize my attribute :
Some time it gives a result as : 'All' (there is no discretization)
and when I turn on my program, I have a problem because that!!
Could you help me please ?many Thanks
It returns "all" because the supervised discretisation could not find
a meaningful split of the attribute with relation to the target class.
To get it to work, either use unsupervised, or use another filter to
delete one value attributes from the datasets, or reprogram the
learner you're using to deal with one value attributes (it really
should be able to handle it). If you need to compare multiple
classifiers on the same dataset, then you could program a "Wrapper"
classifier that does the filtering automatically. I have such a
wrapper I wrote for the "AODE" classifier which auto discretises which
you could modify for your needs. If you want it, let me know.