Exploring the Use of Recurrent Neural Networks for Time Series Forecasting

Recurrent Neural Networks (RNNs) have become competitive forecasting methods, as most notably shown in the winning method of the recent M4 competition. However, established statistical models such as exponential smoothing (ETS) and the autoregressive integrated moving average (ARIMA) gain their popularity not only from their high accuracy, but also because they are suitable for non-expert users in that they are robust, efficient, and automatic. In these areas, RNNs have still a long way to go. We present an extensive empirical study and an open-source software framework of existing RNN architectures for forecasting, and we develop guidelines and best practices for their use. Recurrent neural networks have been effectively used to predict outcomes from irregular time series data in a variety of industries, including medicine, traffic monitoring, environmental monitoring, and human activity detection. The paper focuses on two widely used methods for dealing with irregular time series data: missing value imputation during the data pre-processing stage and algorithm modification to deal with missing values directly during the learning process. Models that can handle problems with irregular time series data are the only ones that are reviewed; a wider variety of models that deal more widely with sequences and regular time series are not included.