https://github.com/nixtla logo
#mlforecast
Title
# mlforecast
m

Matej

10/07/2023, 11:36 AM
Can you fit MLforecast object for only one specific horizon, which is not 1?
Copy code
horizon = 6

fcst.fit(
    df_train, 
    static_features=[], 
    max_horizon=horizon
    )
• in the above code I understand I train models for all horizons=1,2,3,...,6 Q: Is there a way to only train model for horizons = 6, the reason might be quicker iteration of models for a specific horizon and then training all horizons after • (Or this might be achievable simply by appropriate transform, where only lag6 is available and then horizon = 1, is that the correct way to approach this ?) Thanks and have a great weekend.
j

José Morales

10/10/2023, 3:20 PM
Hey. I believe you should use lags>=7, so that the lag7 is the lag1 for your desired target. That'd give you an estimate of how good the model would be
nixtlablack 1
👍 1
2 Views