In terms of the number of rules created, RULES-IS compares favourably with the other algorithms tested, generating fewer rules than even C5.0, which is considered to be amongst one of the best algorithms currently available. However, the accuracy of the four rules formed by RULES-IS is less than the 100% achieved by both C5.0 and RULES-3, but at least equivalent to the accuracy of RULES-4.

When the accuracy of the rules generated by RULES-IS in classifying unseen data is examined it can be seen that of the examples classified an accuracy of 95% was achieved.  This is comparable to the other algorithms lying approximately mid-way between the worst (C5.0) and the best (RULES-3).