| dc.description.abstract | Data mining can be used to discover the useful informations and rules from the tens of millions of data, and it can identify the target customers and make better decisions. Classification is a one of the data mining domain which a very wide range of application technologies, based on available information and categories of property to create a data classification model, and use this model to forecast class of the unclassified data, while the decision tree is the most commonly used to the a classification model, because it generates easy to understand rules, and fast to establishment and simple; the traditional decision tree need to gather a complete historical data, when it grow the decision tree, but the managers hope to make decision early in time.
Therefore, this study given the delay time (time stamp) for each attribute, and build a decision tree TStree which to consider the concept of time stamp, and the decision-makers hope to find the rules, it can be decision-making point early, and the nice rate of the classification accuracy through TStree.
The results show that we gather some information from the complete data set, then we can make decisions early through TStree Algorithm, and the accuracy of TStree algorithm is close to the accuracy of the traditional algorithm (i.e. ID3, C4.5, etc.)
| en_US |