Hi,
NARNET predicts future values of a time series using only its past values. It's simple and works well if the series is not influenced by outside factors. It might not perform well if there are outside influences because it can't use that extra information.
NARXNET, on the other hand, can include outside influences in its predictions, making it better for situations where external factors matter. This makes NARXNET more powerful, but it also requires more data and is a bit more complex to set up.
Choose NARXNET if outside factors are important for your predictions.
You can also go through following related MATLAB Answer:
