https://github.com/nixtla logo
m

Mehmet Can Yıldırım

08/29/2023, 6:43 AM
Hi everyone, I use statsforecast for my time series project that is based for demand forecasting. So, i have a question for you. I am trying to predict next 13 periods. In some series, model gave me good results when compare with test values. However, in some series, i encounter a problem. The problem is to predict same values in every 13 periods. Is there any reason for this? And can i fix this problem via using model parameters etc.? Thank you for interest.
b

Bartosz Bohaterewicz

08/29/2023, 12:48 PM
you get same value each 13'th period (day,month whatever)?
m

Mehmet Can Yıldırım

08/29/2023, 12:54 PM
Yes, its true. My data is weekly and i get same values each of them.
b

Bartosz Bohaterewicz

08/29/2023, 12:54 PM
so each 13'th week?
m

Mehmet Can Yıldırım

08/29/2023, 12:55 PM
No. For example, i get 100 for first week, and then i get also 100 for second week, and for all 13 weeks etc.
b

Bartosz Bohaterewicz

08/29/2023, 12:55 PM
so straight line?
m

Mehmet Can Yıldırım

08/29/2023, 12:55 PM
Yes, that's true
b

Bartosz Bohaterewicz

08/29/2023, 12:55 PM
check your features
m

Mehmet Can Yıldırım

08/29/2023, 12:57 PM
actually, i don't have features, i have only 1 variable that is my target variable
b

Bartosz Bohaterewicz

08/29/2023, 12:58 PM
model wasn't able to capture any significant patterns/variations in your training dataset.
1
which models are u using?
m

Mehmet Can Yıldırım

08/29/2023, 12:58 PM
Oh, i get it now. I did some research. I also encounter this.
Actually i use many models and get many predictions. However, i select the best model when comparing validation set. I can give a model for an example. (DynamicOptimizedTheta)
b

Bartosz Bohaterewicz

08/29/2023, 1:00 PM
add some features and check again 🙂
m

Mehmet Can Yıldırım

08/29/2023, 1:00 PM
Ok, thank you for your answer🙂