在Python中的ARIMA预测

Question

I have a time-series forecasting problem that I am using the statsmodels python package, I applied the ARIMA MODEL, In python sm.tsa.ARIMA(data, (p,1,q)) usually transform the data to the first different, for example if we have a raw data (y1,y2,y3,y4....), first thing ARIMA Find the first difference,(y1-y2,y2-y3,....), so it make the model from this new data (first difference data). my question when I found the model

arma_mod1=sm.tsa.ARIMA(firstdifference, (p,1,q))

I can predict the first difference data as follow

predict_oil =arma_mod11.predict('1980', '2026').

MY QUESTION: How can I predict the future raw data ( the main data not the first difference data) using Arima?

Thanks

Answer 1

The predict method takes an optional parameter named typ which lets you decide whether to have predictions in the original time series or in the differenced one.

You should use

predict_oil =arma_mod11.predict('1980', '2026', typ='levels')

I don't think this will be still helpful for you, but maybe it will be for others.