Market Basket Analysis (MBA) is a popular data mining technique, frequently used by marketing and ecommerce professionals to reveal affinities between individual products or product groupings. The general goal of data mining is to extract interesting correlated information from a large collection of data–for example, millions of supermarket or credit card sales transactions. Market Basket Analysis helps us identify items likely to be purchased together, and association rule mining finds correlations between items in a set of transactions. Marketers may then use these association rules to place correlated products next to each other on store shelves or online so that customers buy more items. Finding frequent sets in mining association rules for Market Basket Analysis is a computationally intensive problem, making it an ideal case for MapReduce.
This chapter provides two Market Basket Analysis solutions:
A MapReduce/Hadoop solution for tuples of order N (where N = 1, 2, 3, ...). This solution just finds the frequent patterns.
A Spark solution, which not only finds frequent patterns, but also generates association rules for them.
This chapter presents a MapReduce solution for data mining analysis to find the most frequently occurring pair of products (order of 1, 2, ...) in baskets at a given supermarket or ecommerce store. Our MapReduce solution is expandable to find the most frequently occurring
N = 1, 2, 3, ...) ...