As it was discussed in class, one may derive a set of classification rules from a decision tree and vice-versa.
computer science
Description
NOTE: If you use a source from the literature make
sure you cite it properly. If you use a
segment already published in the literature, make sure you clearly indicate so
by providing the source and use that segment within “quotes”
and a different font type.
HW#3: (Maximum score = 100
points)
MAIN GOAL: To identify
some key relations between some typical classification approaches and
association rules.
Problem description:
As it was discussed in class, one may derive
a set of classification rules from a decision tree and vice-versa. Association rules, on the other hand, can be
used to express relations between pairs of mutually exclusive sets of attributes. Next, assume we work with datasets defined on
binary variables.
Questions:
Is it possible to use association
rules to derive classification rules? How
can the class attribute be defined in that case? How about use a method that infers classification
rules be used to infer association rules?
Design some algorithms (describe them in pseudocode) and demonstrate
your ideas by using small toy-size datasets with binary attributes. Thus, describe the following cases:
(1) How to infer a
decision tree by using a classification rule inference method.
(2) How to infer classification
rules by using a decision tree method.
(3) How to infer association
rules by using a classification rule method.
(4) How to infer
classification rules by using an association rule method.
(5) Finally, how
can we treat data with non-binary values if we wish to apply an association rule
inference approach?
