Scroll Top
100 Powdermill Road, Suite 108, Acton, MA 01720

Improving Demand Forecast Accuracy with Abnormal Demand During Pandemic like COVID 19

2020 is destined to be different from other years. The COVID-19 pandemic has significantly impacted normal demand patterns in almost all industries. Major examples include:

  • Grocery: The grocery industry saw a massive demand increase during the pandemic periods, due to concerns over grocery availability. If we do not pre-treat the demand history during these periods, it will overestimate the demand for the same period in 2021 and beyond.
  • Transportation: As consumers began working from home and avoiding tourism, the demand for transportation services dropped significantly. If we do not pre-treat such demand history, any statistical forecast engine will underestimate the demand for the same period in later years.

How do we account for the outliers during the pandemic so that we can get a better forecast accuracy in the future years? A simple approach would be to adjust the historical demand item by item during the pandemic periods. Although this method would work, it is neither efficient nor practical: the normal demand for each item would have to be manually “forecasted” during these periods.

A more advanced approach is to automate this process. For example, one can setup a flag to inform the statistical forecast engine to ignore the sales history data during a pandemic period. A good statistical forecast engine should come up with an algorithm to “forecast” the normal demand of an abnormal period. The best approach would be to use seasonal and yearly growth to project the normal demand. For example, if the sales for March 2019 was 100 cases, and the growth year over year was 5%, then a good “forecast” for March 2020 would be 100*(1+5%)=105. This number would replace the actual sales history of March 2020 inside the pandemic period, which will then be used to forecast future demand.

New Horizon Soft has incorporated this approach in its statistical forecast engine. Using Artificial Intelligence to “forecast” the normal demand during these periods, we found that our approach is both extremely efficient and accurate. If you are interested or have any questions, please contact us at [email protected] for more details.