This message was deleted.
# generals
Slackbot
11/14/2022, 4:04 PMThis message was deleted.
m
Max (Nixtla)
11/14/2022, 4:08 PMI think the problem is that you have a unique_id for every timestamp. Try changing the unique_id value to something like “1” or “Series 1" if you are just using 1 series.
Max (Nixtla)
11/14/2022, 4:09 PMStatsForecast can also handle multiple unique_ids automatically. If that is what you are looking for.
Max (Nixtla)
11/14/2022, 4:12 PMAlso, if your data is weekly and you have more than 1 year of obserbations you can set the
season_lengthMax (Nixtla)
11/14/2022, 4:14 PMSo, it would look something like this
Copy code
df_agg2 = df_agg2.assign(unique_id='1')
sf = StatsForecast(df=df_agg2,
models=[AutoARIMA(season_lenght=52)],
freq='W',
n_jobs = -1)
forecasts = sf.forecast(7)
forecasts.head()j
Jose Bordon
11/14/2022, 4:14 PMAh yeah the unique_id all being set to 1 worked
Jose Bordon
11/14/2022, 4:15 PMWhat if the seasonality is more monthly than weekly though?
m
Max (Nixtla)
11/14/2022, 4:15 PMFor setting the optimal season_lenght you can check this article: https://robjhyndman.com/hyndsight/seasonal-periods/
j
Jose Bordon
11/14/2022, 4:15 PMgreat thank you!
m
Max (Nixtla)
11/14/2022, 4:17 PMYou are welcome. Also, feel free to try more models and see what works best.
You just have to change this line:
Copy code
models=[AutoARIMA(season_lenght=52), ETS(season_lenght=52), ... ],Max (Nixtla)
11/14/2022, 4:17 PMHappy to help if you need anything else 🙂