I am trying to replicate results of the Hyndman's R version of Arima function with SARIMAX function in Python, but the coefficients are not matching given the same dataset.
R code:
md1 = Arima(y, xreg = as.matrix(df[,c('x1', 'x2'), drop = False]), order = c(1,0,1), include mean = TRUE)
Python code:
md2 = SARIMAX(y, exog = df[np.array(['x1', 'x2']), trend = 'c', mle_regression = True)