0
$\begingroup$

I am preparing a regression analysis of sales data and advertisement data. The sales data is from Monday to Sunday, but Sundays have a value of 0 because the store is closed. The advertisement data from Monday to Sunday has values way greater than 0 on Sunday. It is for a Marketing Mix Modeling project.

How can I deal with that? Is that called noisy data or something else? Do I just remove the Sunday altogether?

Thank you in advance for any hints how to research that further.

Improve this question
$\endgroup$
4
  • $\begingroup$ Use a dummy variable with value 1 for sunday and 0 for everything else. $\endgroup$
    – user2974951
    Commented Oct 4, 2019 at 12:40
  • $\begingroup$ How is a dummy variable going to help in that case? I don't understand. $\endgroup$
    – Moritz Möller
    Commented Oct 5, 2019 at 16:40
  • $\begingroup$ Ok. I found the answer myself. Basically it is a time series analysis and I just need to deseasonalize it. $\endgroup$
    – Moritz Möller
    Commented Oct 7, 2019 at 5:17
  • $\begingroup$ No, a dummy variable will tell the model to treat sundays differently than the other days. $\endgroup$
    – user2974951
    Commented Oct 7, 2019 at 8:05

1 Answer 1

Reset to default
0
$\begingroup$

Use 6 dummies to deal with day-of-the-week and delete any of these indicators if they are not significant. Introduce Holiday indicators allowing for lead , contemporaneous and lag effects. Look at Simple method of forecasting number of guests given current and historical data for more model development strategies.

Improve this answer
$\endgroup$

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.