ABSTRACT: The Inability to determine better patterns in a group of events that will lead to particular deviations in customers’ behaviour and difficulty in detection of a repetitive approach to a particular sales company or a group of companies in order to identify regularities in the business or in a particular sector of a business as regards to purchasing item at different seasons. which have become a problem in data mining. However, periodic mining patterns from transactional database require an exponential mining space to produce a huge number of patterns, the first priority for a mining algorithm is the efficient discovery of user-interest-based periodic patterns. It is often necessary to mine a limited, interesting representative subset of frequent trends in many real-world scenario. This paperpresents a model for efficient exploration of periodic pattern in big data. Suffix and prefix trees have been used to capture the contents of the database in a very compact way to generate the full set of periodic-frequent patterns in a database for frequency and support thresholds provided by the user. A periodic pattern algorithm was developed to efficiently list all periodic item-sets. Themodel was implemented in Jupyter notebook using python programming language. The results show that some of the patterns discovered in this database are appearing not only frequently within the database but also appearing at regular intervals within the database at minSup 0.1% and maxPrd 10%.
Keywords: Data mining, Support Vector Machine, Logistic Regression, Recognition
| DOI: 10.17148/IJARCCE.2022.11129