The Technics Of Mining Positive And Negative Association Rules Based On Sequential Patterns

With the rapid development of database technology and broad application of databasemanagement system, database mining was born. We can get very important information for sales and production through database mining in certain pattern, so as to improve sales and production efficiency, and reduce costs for maximum commercial benefit.Sequential patterns'mining is an active branch of database mining. Through mining sequential patterns between frequent itemsets, buyer's shopping pattern can be predicted; quantity of sale can be promoted. For example A?B, this rule refers customers to buy merchandise A, then often buy B. Businesses can constitute sales strategy according to this rule to promote sales of the two products. But, sometimes customers buy merchandise A, tend not to buy C, this rule is denoted as A??C, we call it negative association rules based on sequential patterns. In the business of decision-making, the negative association rules based on sequential pattern on how to reduce the negative factors, and increase benefit in the maximum are particularly important.Now, most researches about sequential patterns are concentrated on the positive association rules, negative association rules based on sequential patterns have not yet commenced research. The thesis studied the existing sequential patterns algorithm and negative association rules mining, used the definition of correlation to remove conflicting association rules, and combined these two algorithms, gave an algorithm of mining positive and negative association rules based on sequential patterns. Finally, gave an examples of the application, specified the implementation of the algorithm.