Hierarchical clustering methods for semiconductor manufacturing data

Our implementation of data mining skills for the data of the semiconductor manufacturing process is reported and the possible relationship between machines of the manufacturing process and the yield rates of wafers is discussed. In our implementation, we considered a full manufacturing process as a transactional data and viewed machines of the manufacturing process as objects of the transactional data. Therefore, the sum of transactional data were obtained from the objects constituted by machines. In our algorithm, we employed the hierarchical clustering methods to distinguish groups according to the similarity of objects. That information can be analyzed and be used to schedule the manufacturing process. Hopefully, the obtained information can provide references for improving the yield rates in the manufacturing process.