consultr - classify items using a rule set


consultr [ -f FNS ] [ -t ]


Consultr reads a rule set produced by c4.5rules and uses this to classify items whose description is provided by the user. The rule set read is filestem.rules, where filestem is the file name stem (see c4.5(1)).

Consultr prompts for the value of an attribute when needed. Unknown attribute values are specified as "?".

If the attribute has discrete values, its value can be specified either as


where v is one of the permissible values of the attribute, or as

v1:p1, v2:p2, ... , vn:pn

where the vi's are some or all of the permissible values of the attribute and the pi's are corresponding probabilities. If the sum of the pi's is less than 1, the remaining probability is distributed equally among the unspecified values of the attribute. For example, if attribute A has possible values X, Y and Z, the reply


is the same as

X:0.5, Y:0.25, Z:0.25

If the attribute has real values, its value can be specified either as


where r is a real number, or as

r1 - r2

where r1 and r2 are real numbers. The latter form is taken to mean "somewhere in the interval [r1,r2]".

When all relevant attributes have been determined, consultr will give a class that the item might belong to. The likelihood of the class may be indicated by a probability. For example,

C1 CF = 0.9

means "class C1 with probability 0.9".

At the end of each item, consultr will prompt for retrial of the same item, a new item, or end of session. When retrying an item, the previous values of queried attributes are displayed in square brackets. If the value is unchanged, reply with a carriage return; otherwise, specify the changed value.

Options and their meanings are:

-f filestem Specify the filename stem (default DF)
-t Display the rule set at the start of the consulting session.




c4.5(1), c4.5rules(1)