A novel algorithm for mining maximal frequent gradual patterns.

The problem of mining frequent gradual patterns has received important attention within the data mining community, because it has many applications in many domains, such as economy, health, education, market, bio-informatics and web mining. Algorithms to extract frequent gradual patterns in the large databases are greedy in CPU time and memory space and the number of frequent patterns generated by these algorithms is sometimes too large to be fully exploited within a reasonable timeframe. This raises the problem of improving the performances of these algorithms and the problem of exploiting concise representations of frequent gradual patterns. This paper presents a new maximal frequent gradual pattern mining approach that relies on an in-depth traversing of the search space in the lexicographical order and a reduction of the search space and the computational load of fundamental operations. This approach leads to a new, more efficient algorithm called MSGriteMiner. Complexity analysis, in terms of CPU time and memory usage, and experiments carried out on various well-known databases show that MSGriteMiner is better than the previous algorithms and confirm the interest of the proposed approach.