Performance analysis of neural network architectures for time series forecasting: A comparative study of RNN, LSTM, GRU, and hybrid models

• A Monte Carlo method to assess machine learning time series algorithms is outlined. • Nine 2-hidden-layer algorithms with RNN, LSTM, and GRU structures are evaluated. These are RNN, LSTM, GRU, RNN-LSTM, RNN-GRU, LSTM-RNN, GRU-RNN, LSTM-GRU, GRU-LSTM. • Over a hundred iterations, LSTM performs the best on one time series dataset and LSTM-RNN on the other two datasets. • Although no method is universally optimal, RNN is the fastest among all methods, and LSTM-RNN is generally faster than LSTM-GRU. Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM) networks, and Gated Recurrent Units (GRUs) have gained significant popularity in time series forecasting across diverse domains including healthcare,…

• UT
  Universiti Teknologi Petronas
• UP
  Universitas Pertamina