I am trying to understand what is the best possible way to execute a time series forecast. I am trying to forecast the number of employees that are going to call in sick on a given day. My data has the current features.

  • Employee hometown
  • Job title of employee
  • Seniority of employee
  • Date
  • Day of week
  • Week of Year
  • Sick count

The Employee hometown, job title, and seniority columns are important because sick numbers historically are heavily dependent on these attributes. Also, I would like to execute an algorithm that also puts a heavier weight on 2021 and 2022 sick counts.

Any ideas would be greatly appreciated!

  • $\begingroup$ can you add an example of your dataset? It seems your features defines an individual employee, but your target informs about a number of sick calls instead of a binary yes/no call? To confirm the way you aggregate your data $\endgroup$
    – German C M
    Commented Aug 24, 2022 at 16:39

1 Answer 1


I assume that sick count is a numerical variable. If so, you could pre-process your data in a way to give a higher value to sick counts for the years $2021$ and $2022$. To make an example, if your sick count for the year $2021$ in the original data is $x = 10$ you could do $5 \cdot x = 50$ to give a higher value on such counts. It is not guaranteed that such a trick will work, but at least you could give it a try as it depends on the forecasting algorithm.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.