Weighted Negative Sequential Patterns Algorithm With Multiple Supports

Based on the previous traditional algorithm,data mining has to equate to the data items and no distinction between the important degree of purpose,so it can't dig the knowledge of the more important for the policy makers.It is based on the single support and the users can set their own supports.If the support is setted too high,the algorithm will ignore some important data and set too low that there will be a redundancy phenomenon.This paper is to study mining weighted frequent negative sequence algorithm based on the multiple supports,and the key is to study how to set up reasonable weightand and multiple supports problem to the items to enhance the excavating performance of algorithm in practical application.To set sequence weights is mainly used to reflect the purpose of importance.The weights are given by the user,and to min the sequences which are more important.To select a minimum support degree by the user,and through the formula to calculate the size of the support after the objective weighting,and digging out more useful information for users.Avoiding due to the single support problem ignoring the important sequences with low support or too much useless and frequent sequences to influence the makers who make decision.It is based on the classical negative sequence pattern algorithm Neg–GSP to add weight and multiple supports to the items on the basis of the GSP.Only the support of sequence is bigger than the project's own weighted support,it can be output as frequent sequence.This algorithm is compared with the classical algorithm called Neg-negative sequence algorithm to have the advantage that items are gived weights and multiple minimum supports.The setting of weights can dig out the sequences in the database of low frequency but high important.The setting of multiple minimum supports can avoid the support is set too high and ignoring the valuable sequences under the single support or the support is set too low and producing the redundant sequences.Through the improvement of the two parts,the algorithm was validated by the experiment on the performance and efficiency of mining sequential patterns is superior to Neg-GSP algorithm and it can dig out more sequences in accordance with requirements of users.