Data-driven analysis of usage-feature interactions for new product design

Abstract

Data-driven design has gained much attention with the rise of big data technologies and the availability of user-generated data. Previous research utilizing user data offered various design implications and ideas for new product development. However, most studies primarily focused on product features with little consideration of product usage, a significant factor in new product design. Moreover, while it is important in design practice to prioritize features in terms of usage, the interaction between usage and feature has rarely been investigated. To address the above limitation, this study proposes a new methodology that quantifies the interactions between product features and usages. The method consists of three stages. First, it analyzes customer sentiments toward product features and usages. Second, the method trains a neural network model that predicts customer satisfaction based on these sentiments. Then the method calculates the impact of each input factor by SHapley Additive exPlanations (SHAP). In the final stage, the impact values are further analyzed by a new function, Effect Measurement based on Covariance Analysis (EMCA), to quantify the interactions of feature and usage factors. The proposed methodology was initially tested on synthetic datasets for validation and then applied to real-world datasets. The result provides numeric values for product usage-feature interaction, which can help companies devise proper strategies for new product development.