Based On The Classification Rule Mining Algorithm And Application Of Fp - Growth

This paper is supported by the National Oceanic and public research topic of "the integration, quick access and intelligent analysis techniques of Bohai Sea marine heterogeneous data". When using FP-Growth algorithm to mine large databases, there is the phenomenon that the FP-Growth algorithm will traverse too much nodes. So the improved FP-Growth algorithm is provided not only to reduce the number of nodes of the tree, but also the number of the traversed nodes, and the cost of time and space.Any N-category classification problem can be transformed into a two-class classification problem. So based on the FP-Growth algorithm, a two-class classification algorithm is provided. In order to avoid the step that the original FP-Growth algorithm generates conditional pattern bases, the tree produced by frequent1-itemsets does not include the class attributes, but adds the count of relative support, obtains the conditional tree with classification rules. This not only reduces the size of the tree, but also reduces the space cost and reduces the generation of invalid classification rules.Based on the improved FP-Growth algorithm, take MODIS images as the data source, mine the detection rules of MODIS images, then presents the green tide detection method based on knowledge. Take green tide in May-July in2008in the waters near Qingdao as research objects, conduct the case studies of green tide detection; the result shows that the mined rules can detect the outbreak range of green tide with higher accuracy. This further illustrates the effectiveness of the improved FP-Growth algorithm.