ARIMA Model In loop python

Question

I have an arima Model trained in a dataset of train[:350] test[350:427] values. I Am fitting the model in train values and my (p,d,q) values are (1,1,2). Currently i can predict only one time step at a time. I want to run the model in loop so that everytime it outputs one forecasted value, it is added to the train dataset and the new value is used to predict another new forecasted value. I am unable to understand how to do it and so far this is what i have got.

historical = train['max']
predictions = []

for t in range(len(test)):
    model = ARIMA(historical, order=(1,1,2))
    model_fit = model.fit()
    output = model_fit.forecast(exog=test['max'][t])
    predictions.append(output)
    observed = test['max'][t]
    historical.append(predictions)
    print(len(historical))

Answer 1

You do not need a loop for this, you use the predict method instead:

model_fit.predict(start, end, endog=test["max"], dynamic=True)

In this case ARIMA will add the predicted values to your data in order to make new predictions. If you want to update your data with the real values instead, you set dynamic=False

https://www.statsmodels.org/dev/generated/statsmodels.tsa.arima.model.ARIMAResults.predict.html#statsmodels.tsa.arima.model.ARIMAResults.predict