One effective way of choosing training set and algorithm on building a decision tree for data mining.
Đoàn Văn Ban, Lê Mạnh Thạnh, Lê Văn Tường Lân · Journal of Computer Science and Cybernetics · 2012
Data mining for the purpose of discovering useful implicit information from data warehouse, i.e. knowledge discovery to serve supporting decision making in our activities, has become more and more important.Therefore, there exists a lot of methods and techniques focusing on the studies and applications for data mining and knowledge discovery.Decision tree is known to be one of the effective solutions to describe the characteristics of mined data.Building an effective decision tree depends on the selection of training set.In practice, business data have been stored in multiform and of complexity, which consequently leads to the difficulty in selecting a good sample training set.If an untypical sample of training set is chosen, it will lead to low practicability in the corresponding decision tree.In this article, we have analysed and presented one effective way of choosing sample training set from business database.Based on this, we will apply learning algorithm to build an effective decision tree of high predictability for supporting decision making in data analysis problems.The obtained results show that proposed this method is more efficient.