ARIMA model is used to for time series and auto regressive neural networks can also be used for time series if you account for direct connections between the input layers nodes and the output layer node. The directed connections helps you to account for the non stationary data. However, ensemble of ARIMA and neural network is no the same as a auto regressive neural network. Particular in EM ensemble models work in a different way by combining the outcomes, meaning the predict probabilities for interval targets and voting or averaging the posterior probability for categorical targets.
Here's a related thread you and others chimed in on in the SAS Forecasting and Econometrics: Communty: https://communities.sas.com/t5/SAS-Forecasting-and-Econometrics/Modelling-ARIMA-ANN/m-p/417428#M2883
Anna
Good news: We've extended SAS Hackathon registration until Sept. 12, so you still have time to be part of our biggest event yet – our five-year anniversary!
Use this tutorial as a handy guide to weigh the pros and cons of these commonly used machine learning algorithms.
Find more tutorials on the SAS Users YouTube channel.