This message was deleted.

Hey. The AutoARIMA model takes arrays as input, are you converting your target to an array?

my workflow thus far has been to pass AutoARIMA to StatsForecast like this

a_arima = AutoARIMA(
    season_length = 7,
    trace = True,
    alias = 'auto_arima'
)
sf = StatsForecast(
    models = [
        a_arima
    ],
    freq = 'D',
    verbose = True,  
)

sf.fit(df)

df

has multiple unique_ids

ah, so you're using the StatsForecast class. In that case the best model is found for each unique_id

oh 🙂 . wonderful! Thanks very much for your assistance.