A Sentiment-Based Approach for Innovative Product Sales Forecasting

Purpose: Established information can be detected, obtained, combined, and deployed to solve new problem domains using data analytics technology, which can speed up the innovation process. External information usage makes rigorous contribution to innovation. Therefore, consumer behavior analysis has become an important instrument for an innovation-driven market, and there are various innovative analytical approaches to predict products’ sale performances. Therefore, this chapter identifies the impact of electronic word of mouth (eWOM) attributes to predict user-sensitive product sales performance.Design/methodology/approach: The Valence Aware Dictionary for sentiment Reasoning (VADER) text analyzing method was deployed to properly identify consumer opinions. The prophet forecasting model was developed as a basic forecasting model and improved for accuracy using different classified variables of eWOM.Findings: The relative improvement of the MAPE value was 10%, 13%, and 42% with sentiment, sentiment with volume percentage, and sentiment with volume use as an extra regressor to base model, respectively. The predictive capability of interplay variables performs a more integral role than those regressors used as an individual regressor to the model. A new eWOM combination was identified which is a novel and innovative contribution to literature and practice. This method reveals data that can assist leaders in discovering new ways to be more innovative.