Decision Tree Classification Method And Its Railway Ticket Marketing Analysis

With the development of computer science, more and more original data is collected and stored in computers. The poor ability of managing data makes data rich and knowledge lean. It is for this reason that data mining, to discover useful knowledge from amount of data by uncommon methods, has developed very quickly. Classification is a widely used technology in data mining. There are many algorithms that have been proposed recently, but most of them were memory based and usually assume that the amount of data is not very large. With the larger and larger amount of data, it becomes a challenging problem to find an efficiency classification algorithm that adapts to large database.With rich data hi tram tickets system, how to mine useful knowledge is an important problem. Applying the technology of classification hi train tickets analysis, we construct a new classification method TT_DTC (Decision Tree Classification based on Train Tickets). We apply new classification algorithm SF_DT (Decision Tree Classification Algorithm based on Splitting Files) that bases on splitting files and quantity rules, which aimed at the characters of train tickets. We realize the analysis and prediction about tickets sale and train operation by this method. This method has been used hi train tickets analysis successfully, and provided uncommon patterns and decision information, and accomplished the connection between classification technology and large database.TT_DTC realizes a series of processes including data preprocess, decision tree classification, producing rules and prediction analysis, which based on the data of train tickets and aimed at the characters of tram tickets which have large amount of data and complex attributes. This method can fully meet train tickets analysis hi railway, can efficiently analyze and deal with train tickets and attain decision knowledge which direct the train operation.The algorithm of SF_DT, which bases on the idea of decision tree classification algorithm IDS, use the means of file splitting take the place of the means which bases on memory. It improves the scalability of classification algorithm and can deal with very large database. Moreover this algorithm can produce quantity rules with statistical information and supply the distribution of main class in details. So it can supply more detail information for data analysis.We found a new application background for classification for further research. Moreover, by applying the classification in the train tickets analysis, we provide rich decision information for the management of train operation.