In Search of Actionable Patterns of Lowest Cost – A Scalable Action Graph Method

Abstract

Action Rules benefit its users to achieve their goals by extracting actionable information hidden in the large data. Undertaking such actionable recommendations incur some form of cost to users. The actionable recommendation system fails when the recommended actions are cost wise unendurable or non-profitable and uninteresting to the end user. Finding low cost actionable patterns in larger datasets is a time consuming and requires a scalable approach. In this work, we give a representation for Action Rules as graphs called Action Graphs, which uncovers undiscovered relationship between actionable patterns in the recommended action rules. Also, we define three popular graph algorithms: Dijkstra's Shortest Path algorithm, Breadth First Search algorithm and Depth First Search algorithm to search low cost Action Rules from Action Graphs in the distributed scenario using Spark framework. Upto our knowledge, our Depth First algorithm is the first work to be implemented using Spark framework. We apply the proposed algorithms to three datasets in transportation, medical, and business domains. Results show these domains can benefit from the discovered actionable recommendations of low cost, in time efficient way.