期刊名称：International Journal of Advanced Computer Science and Applications(IJACSA)
印刷版ISSN：2158-107X
电子版ISSN：2156-5570
出版年度：2019
卷号：10
期号：10
页码：61-68
出版社：Science and Information Society (SAI)
摘要：Transactional data streams (TDS) are incremental
in nature thus, the process of mining is complicated. Such
complications arise from challenges such as infinite length,
feature evolution, concept evolution and concept drift. Tracking
concept drift challenge is very difficult, thus very important for
Market Basket Analysis (MBA) applications. Hence, the need for
a strategy to accurately determine the suitability of item pairs
within the available billions of pairs to solve concept drift
chalenge of TDS in MBA. In this work, a Scalable Data Analytics
Market Basket Model (SDAMBM) that handles concept drift
issues in MBA was developed. Transactional data of 1,112,000
were extracted from a grocery store using Extraction,
Transformation and Loading approach and 556,000 instances of
the data were simulated from a cloud database. Calibev function
was used to caliberate the data nodes. Lugui 7.2.9 and
Comprehensive R Archive Network were used for table pivoting
between the simulated data and the data collected. The
SDAMBM was developed using a combination of components
from elixir big data architecture, the research conceptual model
and consumer behavior theories. Toad Modeler was then used to
assemble the model. The SDAMBM was implemented using
Monarch and Tableau to generate insights and data visualization
of the transactions. Intelligent interpreters for auto decision grid,
selectivity mechanism and customer insights were used as metrics
to evaluate the model. The result showed that 79% of the
customers from the customers’ consumption pattern of the
SDAMBM preferred buying snacks and drink as shown in the
visualization report through the SDAMBM visualization
dashboard. Finally, this study provided a data analytics
approach for managing concept drift challenge in customers’
buying pattern. Furthermore, a distinctive model for managing
concept drift was also achieved. It is therefore recommended that
the SDAMBM should be adopted for the enhancement of
customers buying and consumption pattern by business ventures,
organizations and retailers.
关键词：Association rule mining; big data analytics; concept
drift; market basket analysis; transactional data streams
