cancel
Showing results for 
Search instead for 
Did you mean: 

ARIMA Model woes

aesekody
Meteoroid

ARIMA Model woes

I am trying to do a time series forecast and my ARIMA model is running the forecast from 2007 to 2100. my data spans from 2007 to 2023. I don't know what I am doing wrong.

 

7 REPLIES 7
aesekody
Meteoroid

ARIMA Model woes

I am trying to do a time series forecast and my ARIMA model is running the forecast from 2007 to 2100. my data spans from 2007 to 2023. I don't know what I am doing wrong.

 

alexnajm
19 - Altair

Re: ARIMA Model woes

What is your ARIMA tool configuration?

aesekody
Meteoroid

Re: ARIMA Model woes

I used the standard configuration

aesekody_0-1680803401316.png

aesekody_1-1680803463246.png

 

 

aesekody_2-1680803510928.png

 

aesekody_3-1680803556126.png

 

 

alexnajm
19 - Altair

Re: ARIMA Model woes

It looks like you are doing options beyond the standard configuration since you are customizing the parameters in the "Model customization (optional)" part! 

 

Each line of your data is a distinct week between 2007 and 2023? Can you send a screenshot of what that looks like?

aesekody
Meteoroid

Re: ARIMA Model woes

aesekody_0-1680804999778.png

 

alexnajm
19 - Altair

Re: ARIMA Model woes

@aesekody can you provide a screenshot of the data, not the output?

Highlighted
aesekody
Meteoroid

Re: ARIMA Model woes

aesekody_0-1680856464366.png

The dates, and counts are random. sometimes daily, and in some cases weekly.

alexnajm
19 - Altair

Re: ARIMA Model woes

Good to know - then I would use some sort of logic to aggregate so that the data is at the weekly level. With differing levels of time, it's likely to result in odd results.

 

One possible way is to identify the Sunday for each date and use that to group by while summing the numeric field. Then your weekly breakdown should be better!