Pre-launch new product demand forecasting using the Bass model: A statistical and machine learning-based approach

• Published in: Technological forecasting & social change 2014-07, Vol.86 (-), p.49-64
• Main Authors: Lee, Hakyeon, Kim, Sang Gook, Park, Hyun-woo, Kang, Pilsung
• Format: Article
• Language: English
• Subjects: Algorithms; Artificial intelligence; Bass model; Diffusion; Ensemble; Forecasting; Forecasting techniques; Machine learning; Machinery; Multivariate linear regression; Power; Pre-launch forecasting; Prediction; Product introduction; Regression analysis; Studies; Television
• ISSN: 0040-1625; 1873-5509
• DOI: 10.1016/j.techfore.2013.08.020

This study proposes a novel approach to the pre-launch forecasting of new product demand based on the Bass model and statistical and machine learning algorithms. The Bass model is used to explain the diffusion process of products while statistical and machine learning algorithms are employed to predict two Bass model parameters prior to launch. Initially, two types of databases (DBs) are constructed: a product attribute DB and a product diffusion DB. Taking the former as inputs and the latter as outputs, single prediction models are developed using six regression algorithms, on the basis of which an ensemble prediction model is constructed in order to enhance predictive power. The experimental validation shows that most single prediction models outperform the conventional analogical method and that the ensemble model improves prediction accuracy further. Based on the developed models, an illustrative example of 3D TV is provided. •We propose a novel approach to the pre-launch forecasting of new product demand.•Product attribute DB and diffusion DB are constructed for existing 87 products.•Six regression algorithms are used to predict two Bass model parameters.•Ensemble prediction models are constructed in order to enhance predictive power.•Demand forecasting of 3D TV is provided as an illustrative example