Digital transformation for cost estimation system via meta-learning and an empirical study in aerospace industry

• Published in: Computers & industrial engineering 2023-10, Vol.184, p.109558, Article 109558
• Main Authors: Van Nguyen, Tran Hong, Huang, Pei-Min, Chien, Chen-Fu, Chang, Chung-Kai
• Format: Article
• Language: English
• Subjects: Aerospace manufacturing; Artificial intelligence; Cost estimation; Decision support system; Digital transformation
• ISSN: 0360-8352; 1879-0550
• DOI: 10.1016/j.cie.2023.109558

By applying the UNISON, this study aims to develop a framework for a digital cost estimation system by integrating data-driven methodologies, search engines for market information and domain knowledge to improve decision quality and efficiently utilize human resources in the context of digital transformation and co-value creation procedure. [Display omitted] •Our research integrates document data into a cost estimation system that uses real-time market data updates.•Meta-Learning and Bayesian Hyperparameter-tuning are applied to improve predictive model accuracy and faster training process.•The proposed flowchart reduces human errors, enhances transparency, standardizes decision-making process, and enables automation.•Case study demonstrates a successful digital transformation, resulting in improvement of productivity, and decision quality.•The proposed approach is effectively implemented in an aerospace manufacturing company. It is increasingly challenging to estimate product cost to optimize the quotation decision for aerospace manufacturing companies owing to the diversity of designations, intermittent demands, complicated decision procedure between different functional departments, and knowledge gaps among the involved decision makers. There is a need of effective solutions to support digital transformation of manual approach in which it can expedite and enhance the accuracy of cost estimation by automating specific tasks and streamlining decision processes. Although a number of studies have been done to enhance the performance of forecasting models, little research has been done to address the interrelations between cost estimation model and the associated decisions for quotation. Focusing on realistic needs, this study aims to develop a digital cost estimation system by integrating data-driven methodologies, search engines, and rule-based decision mechanism based on domain knowledge for improving the accuracy of cost estimation and the effectiveness of quotation for revenue management. An empirical study was conducted in a global aerospace manufacturer in Taiwan for validation. The results have shown the practical viability for the proposed framework with better performance than conventional approaches. The developed solution has been implemented.