Based on my augmented-dickey fuller test, the original data seems stationary
To determine my p and q values, I plotted my pacf and acf plots but the first value to go above the confidence band is around ~92 for both. I know this has a quarterly trend (3 months) so it makes sense that I see this pattern, but it's taking far too long to run in my model and I wanted to know if I was analyzing the plots properly.
Here is the code I'm using to implement the model:
model = sm.tsa.statespace.SARIMAX(train, order = (92,0,92), enforce_stationarity = False, enforce_invertibility = False) model_aic = model.fit() print(model_aic.summary().tables)