چکیده:
Association rules are one of the most popular models in the field of data mining. In classic association rule mining algorithms, such as Apriori, a single minimum support range for the rule is used; while in newer approaches that have tried to improve classic algorithms, such as MS-Apriori, a multiple minimum support range for the rule is used, in both cases, the user is obliged to determine the minimum support range of the rule. Consider the user intends to apply the Apriori algorithm to a database with millions of transactions; the user certainly cannot have the necessary knowledge about all transactions in the database, therefore they cannot determine a suitable threshold. In this research, for the first time, using fuzzified data and averaging techniques, a method has been presented in which the Apriori algorithm automatically determines the minimum support range of the rule. Simulated results on a real sample showed that this approach has better performance than the classic Apriori algorithm.
خلاصه ماشینی:
Automatic Determination of Minimum Support Range in Fuzzy Environment to Improve Association Rule Extraction Using Apriori Algorithm Heydar Jafarzadeh 1, Chamran Asgari 2, Amir Amiri 3 Association rules (community rules) are considered one of the most popular models in the field of data mining.
In this research, for the first time, using fuzzified data and averaging technique, a method has been presented in which the Apriori algorithm automatically determines the minimum support range of the rule.
Empirical Background The Apriori algorithm is the most classic and popular algorithm for extracting association rules from databases, which was developed by Agrawal and Srikant in 1994 to extract association rules from frequent data item sets in transactional databases using two pre-defined threshold criteria called the support range of the rule and the confidence level of the rule.
Chen, Hong, and Tseng (2009) tried to improve their previous work based on a fuzzy genetic data mining algorithm for extracting the minimum support range and membership functions for data items in a quantitative database.
Research Methodology In general, the proposed approach to improve the Apriori algorithm is implemented in two stages: • Converting deterministic data to fuzzy using the FCM clustering algorithm; • Apriori approach with improvement in how to determine the minimum support range automatically, which uses the fuzzy dataset to discover frequent patterns and extract association rules.
Using the Apriori algorithm based on the proposed method to determine the minimum support range for the purpose of discovering frequent patterns and extracting association rules, and finally, storing the rules under the name KB.
(
, 
, 
)
