Amazon Product Brand Analysis Framework Using Apache Spark on Real-Time Consumer’s Perception

Abstract

There is enormous growth of Internet users of today’s data science world. Consumers usually prefer buying and finding items online through leading sites like Amazon in the busy hi-tech schedule. They usually read other customer’s feedback for product brand and compare it accordingly based on different criteria. Brand building and positioning involve concerted efforts of several moving parts to establish a market image that moves the company forward. In this paper, we introduce a framework for cohort analysis which highlights activities of people over a period of time. Generally, marketer uses month wise analysis for prediction of product brand sale performance-based customers review. The outcome of the analysis can be used to check the patterns or changes in customer behavior and views toward the product and why there is a change by comparing result. Apache Spark is a flexible in-memory framework for distributed computing and can also handle structured data with Hive. Apache Spark has access to HDFS and can run on both existing Hadoop clusters and new ones. It can handle much more complex data structures than MapReduce, especially compared to MapReduce.