We have described an algorithm (AFT) for decomposing ERDs into various modules / clusters, forming various tables equivalent to number of clusters generated and forecasting based on available raw data and generated clusters. Unlike earlier efforts, our algorithm clusters not only the entities but also the relationships involved. While designing this tool we have thought of its usability to all team members, application domain and extensibility to today’s distributed systems as well. As of now this algorithm is fully automatic, works only on data available in the form of entities, relationships, tuples, fields, feature vector etc. It identifies suitable entity and relationship clusters without any further human (subjective) intervention. The next phase of this algorithm (AFT) will involve human intervention also. One of the corporate companies have given one important feedback. They have one separate department in their organization made up of very knowledgeable and experienced team members only. Their aim is to learn customer thoroughly and think from customer’s business point of view. This will give a value added services to customer. To achieve the same we may add, compare / contrast “expert’s opinion” also. We have discussed how our algorithm AFT, fulfills a comprehensive set of criteria for a good decomposition of ERDs. Our algorithm produces a more cohesive set of clusters while keeping inter-cluster coupling small. Our solution also offers a higher degree of modularity than that offered by other algorithms’ solutions. While our algorithm produces very good solutions, it cannot guarantee their global optimality. Our forecasting module of this proposed algorithm AFT will definitely prove to be most useful and suitable for all corporate teams, thereby saving their precious time which can be utilized on some other important chores.

Databases, Information systems, Analysis and designs, Planning, Decision analysis, forecasting, clustering, entities, ERD, ERM