In recent years industries have increasingly used customer relationship labor in management as a tool to improve their position in the marketplace. This research involves using a data warehouse, decision-tree-based data mining, and neural network pattern classification analysis to isolate the causes of non-conformity in IC packaging. The correctness of the classification analysis produced using the two methods is compared. Our objective is to establish an information analysis system, that is able to quickly identify the causes of problems thereby reducing the time taken to solve quality-related problems. It is shown that predictions made about the target group using decision tree analysis are more accurate than those made by neural network classification, indicating that decision tree analysis is an effective means of classification analysis of a company's quality problems.