Abstract
In order to implement intermediate and final evaluations in product development more efficiently and objectively, a relational model of each evaluation indicator was constructed based on mathematical evidences. First, the relationship between Key Goal Indicators (KGIs), such as three-year sales, market share, and intellectual property creation in product development, and Key Performance Indicators (KPIs) used for intermediate evaluation of KGIs were modeled using Bayesian network analysis based on approximately 1,000 data points. Furthermore, the clusters generated due to the differences in business type and annual sales of the enterprises that conducted product development were converted into nodes and incorporated into the model, which made it possible to grasp the presence or absence of the effects in terms of the correlation generated between indicators, attributable to the characteristics of the enterprise. The model constructed in this way revealed that, irrespective of enterprise attributes, the status of human resource development affects ”three-year sales” of a developed product, while the failure rate affect ”three-year market share”. Besides this, based on this model, key management points in product development were extracted, along with the directionality of examination. Moreover, the validity of these results was confirmed through comparison using factor analysis and a survey of related studies. In the future, we will further refine the detection processing of the influence, especially in terms of enterprise attributes in the proposed method, attempt to improve it through demonstration experiments, and implement it in real society as a data-driven management support engine in a collaborative platform that promotes various innovations.
