Semiconductor wafer fabrication companies rely on continuously improving cycle time for agile manufacturing to maintain their competitive advantages. This study has illustrated how to apply data mining techniques, Back-Propagation Neural Networks (BPNN), to optimize the allocation of production resources in semiconductor wafer fabrication. The results in the empirical study have demonstrated significant benefit for cycle time reduction which contributed 4.36% and 2.49% from eliminating “COV of machine group loading” and “COV of operator loading” respectively.