I am doing a rolling forecast by building multiple sarima models in loop. When I run the code, it works well for many input but breaks down at some particular point providing LinAlgError: Schur decomposition solver error. The code providing Error is:
model_sarima_oneyr = model_sarima_oneyr.fit(training_set_y, seasonal=True, m=7)
I want to run code while skipping the error line (maybe just ignore the error), or if there is a matrix inversion error go to the alternate model, in which this kind of issue does not occur.
The error most likely has occurred due to method='lbfgsb'; could I use another method which does not require matrix inversion and so no error?