Proposed Association Rules Mining Algorithm for Sensors Data Streams
Waheed A.K. Salman, Sattar B. Sadkhan · 2021
Smart environments data mining that depends on sensors is considered one of the most important and recent types of data mining in recent times, but at the same time this type of mining has several challenges, including that the data coming out of the sensors is streaming data and not static in addition to that it contains a time factor in most cases, which It is neglected in mining and data analysis operations. One of the most important tools of data mining is association rules mining, which are considered among the tools of decision-making systems . In this paper we suggest a proposed algorithm called SDT-B-ARM (Data Stream Time Based Association Rules Mining) that deals with time during the mining process and finding frequent itemsets and then finding of association rules from streaming data and all of this is done without the need to store the streaming data. The proposed algorithm needs less storage space, less computational complexity, less time required than previous related algorithms, and more efficiency to obtain strong association rules, in addition to taking time into consideration during the process of finding frequent itemsets. In the proposed approach we can obtain association times which can be useful in applications where time for correlation is important and critical. In addition, the proposed algorithm can be used in a distributed or parallel system.