I've been trying to figure out a solution to this problem for the past couple of weeks and after all my efforts I realized this is a very niche problem. I'm trying to forecast data for event fundraising, however, our donations work on a fiscal year basis so people can donate to next year's event in the previous fiscal year. This results in donations coming in on the same day for different fiscal years, sample below:

enter image description here

Now you can see how this is an issue when working with time series because I've learned that the index has to be a unique Date value. I've been trying to implement a SARIMA model due to the clear seasonal behaviors of my data but no matter what I try it all leads back to having to have a unique datetime index. I tried to make an index out of my Fiscal Year and Days Out (Days before event day) columns but the model says the index is not compatible and runs the model without it giving poor forecast results.

I looked into STL models today and it was similar issue where I have to have a unique daily index.

I tried to move away from Time Series and just do a standard Linear regression model but the R-Square and MSE I got from it made me cry.

Please advise, I have to cook something together by the end of next week. Real World data is pain.


1 Answer 1


Not quite sure what is the horizon you need to forecast, but would it be possible to use the very early donations you've already received, and then look into historical pattern how many % of the total donations is usually received by this day and then just add the missing ones? Sometimes simple ways can be effective as well.

  • $\begingroup$ Your answer could be improved with additional supporting information. Please edit to add further details, such as citations or documentation, so that others can confirm that your answer is correct. You can find more information on how to write good answers in the help center. $\endgroup$
    – Community Bot
    Commented May 13 at 4:27

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.