I am using exponential smoothing and using tableau for forecasting. The first model I included trend and removed seasonality and it predicted the number of cases going up but the quality I got according to tables was "Ok" (Tableau has 3 ranges to describe the quality of a model -Poor, OK and Good). The second model I removed trend and added seasonality to it, This showed the number of cases going down and the quality of the model was good according to tableau. Which model should I be using here, My professor says it's flattening the curve that's why it's going down in the second model while my teammates argue the first model should be the correct one. Model with the trend and without seasonality


1 Answer 1


I will suggest, first take some training data and make both models on that and compare the result on test data e.g

If you have data from 1st Jan 2020 to 28 Apr 2020.

Give the model, data from 1st Jan 2020 to 31st March(Or you can decide another range), make the model and forecast for April month.

And compare the results for both the models on April month(1st April to 28th April).

In this way, you will be able to argue better with results.

  • $\begingroup$ Great, I'll do that. I am using tableau forecasting at the moment where you have to set your own measures and it does the forecasting for you. Would you recommend using the forecasting tool I used in tableau, If yes can I make a judgement looking at the quality metrics like MAPE,MASE. etc and select a model between the both l $\endgroup$
    – user96058
    Apr 29, 2020 at 15:55

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.