In 2015, our team proposed a new development methodology, which we named the Causality Search T-Method (CS-T Method). This method makes it possible to solve the intrinsic limitation of target characteristics-based Parameter Design. Specifically, target characteristics-based Parameter Design is in essence a black-box method, which makes it difficult to obtain information on the mechanisms of quality improvement. The first aim of the CS-T Method is to determine the causal relationships between the target characteristics and multiple candidate “Effective-Explanation Factors” (EEF) such as physical properties, sensing data. The second aim is to improve the efficiency. Through a case study, our team demonstrated that it is possible to determine the causal relationships with significantly fewer experiments.
We propose an extension of the CS-T Method, one which incorporates Graphical Modeling (GM), which we have named the CS-TG Method. Unlike conventional GM, which performs the analysis on the entire pool of candidate EEFs, CS-TG method allows the GM analysis to focus on the limited set of factors that were identified by the original CS-T as having a causal relationship with the target characteristics. In doing so, the new method is able to establish the causal relationships between each of the individual EEFs with fewer experiments.