CODEWITHSUNDEEP
pythontime-seriesarima
Sri2110
Sri2110

Reputation: 335

auto_arima returning the best model as SARIMAX even when time series is stationary and has no seasonal component in Python

I have a time series dataset having frequency as daily. I have checked that my dataset is stationary using augmented dickey-fuller test.

After which, when i am trying to determine the values of p,d,q using:

from pmdarima import auto_arima
stepwise_fit = auto_arima(df2['Births'],start_p=0,max_p=6, start_q=0, max_q=3, seasonal=False,trace=True)

Moreover, I have mentioned seasonal=False in auto_arima argument, but when i did:

stepwise_fit.summary()

Its returning:

SARIMAX Results
Dep. Variable:  y   No. Observations:   365
Model:  SARIMAX(1, 1, 1)    Log Likelihood  -1226.077
Date:   Mon, 17 Feb 2020    AIC 2460.154
Time:   20:02:17    BIC 2475.743
Sample: 0   HQIC    2466.350
- 365       
Covariance Type:    opg     
             coef   std err   z     P>|z|   [0.025  0.975]
intercept   0.0132  0.014   0.975   0.330   -0.013  0.040
ar.L1       0.1299  0.059   2.217   0.027   0.015   0.245
ma.L1      -0.9694  0.016   -62.235 0.000   -1.000  -0.939
sigma2      48.9989 3.432   14.279  0.000   42.273  55.725
Ljung-Box (Q):  36.69   Jarque-Bera (JB):   26.17
Prob(Q):        0.62    Prob(JB):   0.00
Heteroskedasticity (H): 0.97    Skew:   0.58
Prob(H) (two-sided):    0.85    Kurtosis:   3.62

We can see, its returning Model: SARIMAX(1, 1, 1). What can we infer from it? Any suggestion is helpful, or if i am missing something.

Upvotes: 3

Views: 1557

Answers (1)

Sri2110
Sri2110

Reputation: 335

I found the reason its showing SARIMAX(1, 1, 1). It simply means ARIMA only because the format for SARIMAX is basically SARIMAX(p,d,q)(P,D,Q) where P,D,Q are seasonal parameters, so, in our case SARIMAX(1,1,1)(0,0,0) seasonal components are zero only.

Upvotes: 2

Related Questions