E-commerce platforms have heavily relied on predictive machine learning models to leverage the massive data generated in daily operations. However, by altering online customer purchasing behavior, Covid-19 has distorted sales prediction models used by sellers and e-commerce platforms, which may lead them to inaccurate strategic decisions. Using a dataset comprised of electronics products from Amazon, the preliminary results shows that the importance of several predictors of online sales has changed after the beginning of the pandemic. In particular, the importance of negative related factors was found to have significantly increased after the start of Covid-19. Furthermore, adding aspect-based sentiments was found to significantly improve sales forecasting especially during the period after the beginning of Covid-19. The study contributes to the literature evaluating the effects of Covid-19 on e-commerce by providing an in-depth understanding of these effects from an unexplored perspective of prediction models.
Sindihebura, Tanguy Tresor; Pu, Xiaodie; and Chen, Jin, "Predictors of Sales and the Covid-19 Disruption: Evidence from an Online Marketplace" (2022). PACIS 2022 Proceedings. 34.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.