A Deep Learning Model to Predicting Changes in Consumer Attributes for New Line-extended Products
This work addresses a domain-specific problem for marketers by providing a tool to predict consumer attribute changes, helping avoid cannibalization and design effective product line marketing strategies, though it appears incremental as it applies a novel deep learning method to an existing marketing challenge.
The paper tackles the problem of predicting changes in consumer attributes for new line-extended products to support marketing decisions, using a Conditional Tabular Variational Auto-Encoder (CTVAE) that generates synthetic data from tabular consumer and product data, with experimental results showing superior prediction performance compared to existing models.
Product line extension is a marketing strategy that enhances a company's sphere of influence. Because excessive line extensions disrupt brand image, only appropriate line extensions based on consumer needs are desirable. Marketers should know the key consumer attributes of the primary customers for new line-extended products before companies enter the market. This paper describes a method for predicting changes in consumer attributes for new line-extended products using a novel deep learning model. The proposed model, Conditional Tabular Variational Auto-Encoder (CTVAE), generates synthetic data from large-scale tabular data of consumers and products. It can provide various implications about effective product line marketing for marketers. The experimental results demonstrate that the CTVAE offers superior prediction performance than existing models. We indicate implications for new products that change containers or flavors for effective product line marketing. The proposed approach has the potential to contribute to avoiding cannibalization and to designing product images and marketing strategies.