Current Proceedings on Technology

Yazarlar: Güncel Sariman, Ecir Uğur Küçüksille

Konular:-

Anahtar Kelimeler:Data Mining,Association Rules,Parallel Programming,Apriori Algorithm

Özet: Increase of digital with data each passing day more and more also stored in large databases, over time led to the need to benefit from the most efficient way of these data. Association rules which is used in data mining, frequently extract meaningful patterns from large datasets. The status of person who purchased X and also Y, means association rules in market basket analysis. However, large data sets cause to time consuming analysis while extracting these rules. With the increasing number of cores on computers, faster and more efficient data mining analysis can be performed. In this study, association rules extracted with parallel programming techniques with the use of bakery data set from University of Calpoly. An analysis carried out much more quickly and efficiently on multiple threads by using parallel library presented with microsoft.net framework 4.0. Data set pre-processed on database and apriori algorithm used with the developed application interface. Parallel and serial analysis were made separately and double association rules were extracted from data set with user defined support, confidence values and number of cores used in the analysis. In this study parallel and serial analysis times compared and emphasized that shorter analysis can be carried out with parallel programming.

---