Pragmatic Approach to Association Rule Learning in Real-World Scenarios

We present a pragmatic approach for designing an efficient tool for extracting knowledge from customer data in the retail industry, e.g. market basket analysis. Association rule learning is an established topic within data mining and knowledge discovery with a large interest from the business intelligence community. With a focus on properties from a real-world environment and with an aim to get customer insights on a cross-hierarchy level, we have chosen to build upon the common Apriori algorithm. This algorithm has been optimized for the chosen real-world environment and adapted for implementation on commonly available computing platforms and workstations using the Microsoft .net framework. Several parallelization strategies have been developed and experimental results indicate that a significant speed-up is possible and that the tool can be utilized for producing valuable information.