Abstract
The COVID-19 pandemic had a significant impact on consumer purchasing behavior, particularly in the food retail sector. In response to this socioeconomic crisis, food retailers adjusted their strategies to align with new consumer preferences. During the pandemic, sweet snacks sales grew globally as it satisfied the desire for snacking for many families confined at home. Detecting the impact of COVID-19 on sales involves comparing pre-pandemic and post-pandemic sales trends to identify deviations attributed to the crisis. This study examines the sales evolution of sweet snacks at two major Portuguese retail chains from January 2018 to June 2023, split into pre-crisis, crisis, and post-crisis periods. The primary goal is to infer and compare forecasting models using ARIMA and Prophet time series models to assess consumer preference changes and, additionally, predict post-crisis future sales. The findings reveal a break in consumption patterns between periods. In the pre-crisis period, sales progressively increased until the lockdown, declining in the crisis period. At the end of the crisis, consumption patterns normalized, but post-crisis, retailers diverged due to their adaptability to new trends. ARIMA models provided better overall accuracy in predicting stable future sales, while Prophet models delivered more precise forecasts in the post-crisis sales patterns.
Recommended Citation
Barradas, Sara and Martins, Patrícia, "Competing Forecasting Models to Study Crisis Periods: The Case of Sweet Snack Sales" (2024). MCIS 2024 Proceedings. 43.
https://aisel.aisnet.org/mcis2024/43