In the age of the digital economy, social media, forums and other online platforms have played active parts in our daily activities. The amount of data digitized and recorded in these platforms have surged exponentially. Many believed that this underexplored unstructured data sources have huge potential in offering insights to policy makers and companies. This paper aims to propose a hybrid approach using inductive and deductive reasoning to identify motivational factors to use e-cigarettes for predictive modelling. A total of 790 comments and discussions relevant to e-cigarette use and motivations to use e-cigarette were scraped and stored from online forums like Reddit, Vapingunderground and e-cigarette-forum. A series of text analytics were conducted on the text corpus and the cluster analysis enabled us to build a predictive model. Using Bayesian Structural Equation Modelling, we concluded that the constructs derived by clustering, i.e. Cost and Convenience and Enjoyment, have significant associations with smokers trying to quit smoking. While health-related issues were inherent to the notion of quitting smoking, enjoyment, cost and convenience were motivational factors which will generate favourable response towards quitting smoking. The findings showed encouraging results from a methodological standpoint and offered insights to policy makers and companies on health-related issues pertaining to the use of e-cigarettes.
Khong, K W.; Tan, S L.; Teng, S; and Ong, F S., "Predictive Modelling Using Unstructured Data From Online Forums: A Case Study on E-cigarette Users" (2018). CONF-IRM 2018 Proceedings. 24.