Time series forecasting using ARIMA and Recurrent Neural Net with LSTM network

Abstract: Autoregressive integrated moving average (ARIMA) or Box-Jenkins Method is a popular linear model for time series forecasting over the decade. Recent research has shown that the use of artificial neural net improves the accuracy of forecasting to large extent. We are proposing the solution in order to extract both Linear and Non-Linear components in data. In this paper, we propose a solution to predict highly accurate results using an aggregation of ARIMA and ANN (Recurrent neural net) to extract Linear and Non-Linear Component of data respectively. Keywords: ARIMA, Box–Jenkins methodology, artificial neural networks, Time series forecasting, recurrent neural net, combined forecasting, long short-term memory.