I have sales training data from 2019-06 to 2020-06 and I have to predict sales from 2020-06 to today. The pandemic had a great impact on sales. Are there methods to make my ML forecasting models more robust? Here are some ideas:

  • Add a binary variable, pandemic, being 1 when date > 2020-03 and 0 otherwise.
  • Alternatively, define two variables hard_pandemic and soft_pandemic describing respectively the data between March/April 2020 and data after May 2020.
  • Weigh the pandemic data more, since they describe better the current situation.

Are there drawbacks to these methods? Are there alternative ways to improve this?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.