Open Access
Subscription Access
Open Access
Subscription Access
TARPIN:Discovering Temporal Association Rules Using P-Tree Based Incremental Algorithm
Subscribe/Renew Journal
Association rule mining is one of very popular data mining method and number of organizations uses this technique to find the frequent item-sets of products to improve the benefits of organizations. There are number of available algorithms for association rule mining which takes multiple scans of database. The complexities of association rule algorithms primarily depends on number of database scan, so by reducing the number of database scans one can improve the time complexity of these algorithms. The purpose of this proposed algorithm is to reduce the number of database scans for discovering the temporal association rules by applying P-Tree algorithm for temporal association rules which takes just one scan of database to find out the association rules.
A new pattern tree algorithm for mining temporal association rules in databases is introduced. This algorithm uses P-Tree (Pattern-Trees) structures for finding temporal association rules in databases. According to different time periods associated with transactions in temporal databases, it will initiate the number of P-Trees and according to time information in transactions it inserts the transactions in created appropriate trees, then using P-Tree association rule mining algorithm it finds out the frequent sets in this P-Tree and then these frequent items are merged with different time periods which will give the association rules with valid time periods. The proposed algorithm is divided in two phases in first phase all item within the transactions are inserted in different P-Trees on which the frequent item-sets are taken out and in second (merge phase) these frequent items are merged and time associates with these items are in listed which indicates that these frequent items are frequent in this time periods. Algorithm is implemented in C++ under Linux platform and evaluated results are compared with existing popular algorithm PPM (Progressive Partition Miner) for discovering temporal association rule.
A new pattern tree algorithm for mining temporal association rules in databases is introduced. This algorithm uses P-Tree (Pattern-Trees) structures for finding temporal association rules in databases. According to different time periods associated with transactions in temporal databases, it will initiate the number of P-Trees and according to time information in transactions it inserts the transactions in created appropriate trees, then using P-Tree association rule mining algorithm it finds out the frequent sets in this P-Tree and then these frequent items are merged with different time periods which will give the association rules with valid time periods. The proposed algorithm is divided in two phases in first phase all item within the transactions are inserted in different P-Trees on which the frequent item-sets are taken out and in second (merge phase) these frequent items are merged and time associates with these items are in listed which indicates that these frequent items are frequent in this time periods. Algorithm is implemented in C++ under Linux platform and evaluated results are compared with existing popular algorithm PPM (Progressive Partition Miner) for discovering temporal association rule.
Keywords
Temporal Association Rules, Temporal Data Mining, P-Tree, Incremental Data Mining.
User
Subscription
Login to verify subscription
Font Size
Information
Abstract Views: 257
PDF Views: 2