A Round Trip Pattern for Building Decision Trees
Liang Jia, Liuhong Yan · International Journal of Database Theory and Application · 2016
This paper introduces a recursive procedure named round trip pattern for generating decision trees based on alterable decision tables. The pattern is mathematically described by two components representing building tree based on table and altering table based on data from application adapting tree respectively. This cyclic pattern starts at creating a decision tree from initial table altered according to Bayes theorem combined with Pareto values of objective functions. The functions reflect time and spacial complexities of decision trees. After the initial tree is implemented by application, relevant system data is collected and analyzed to alter table for building new tree. This cycle continues until system performance is satisfactory. A small business project is designed and maintained based on round trip pattern. The system statues are retrieved from its database and analyzed in details. The analyses of data indicate system performance stays satisfactorily stable.