Rule induction is one of the most common forms of knowledge acquisition from databases. Most of the existing algorithms search for all possible interesting patterns in a data set so, they can be overwhelming with the large number of rules that can be generated, specially in real values database. Furthermore, the output rules are independent so, many may contradict; each other and they may not cover all possible situations. Therefore, this paper presents an intelligent data-mining algorithm for extracting accurate and comprehensible rules from a given database. The algorithm treats the real values in a given database via a suitable membership function. It corporates a heuristic function which control the rules induction according to three parameters. These parameters are depth of search level, credit, and shoring level. The proposed algorithm is very competitive with the other algorithms such as ID3 and RATIO |
