Temporal analysis of consumer preferences through natural language processing

• Main Author: Bhayangkara, Andhika Satriya
• Other Authors: Chen Songlin
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Industrial engineering::Engineering management
• Online Access: https://hdl.handle.net/10356/158092
• Institution: Nanyang Technological University

Data availability has increased significantly in many shapes and forms, including online customer reviews. This review data has great potential to be a source of information for manufacturers to understand consumer needs and preferences. Nonetheless, customer reviews are still heavily underutilized by manufacturers due to the unstructured nature of the data. Additionally, it is challenging to quantify customer preferences as they are rarely static and evolve rapidly. This study proposes a framework to track and analyze how customer preferences evolve with respect to time through big data analytics with Natural Language Processing. The framework leverages on word frequency analysis and sentiment analysis to derive product feature importance and performance. The following results will be aggregated in a sentiment-based importance-performance analysis model to understand which product features are most in need of improvement. Based on this knowledge, a product improvement strategy can be derived by also considering the most satisfactory product specifications in the market. A case study was performed on smartphone reviews from amazon.com to demonstrate the framework. The proposed framework can be utilized for companies to understand customer preferences which may facilitate companies' decision-making process. Through clear customer insight metrics, more informed product development-related decisions can be made.