Abstract: To adapt to the market changes, a supply chain should be reconfigured from time to time. The supply chain configuration is discussed in this paper and the study is conducted based on the simplified twotier supply chain. By using SPSS Clementine software, a data flow model is built for the adaptive supply chain configuration (ASCC). By using data that include information of purchase quantity, lead time, and price of order as input of the model for training, a rule set for best supplier selection can be obtained. The ASCC model is evaluated by using gains chart and lift chart, result shows that the model is effective. A set of virtual order data is used to test the ASCC model. After training, a supplier is selected with a satisfactory confidence level.

Key words:  adaptive supply chain; , machine learning； supply chain management.