Timeseries Forecasting with RNNs

A weather forecasting experiment using three recurrent neural network (RNN) architectures: a basic RNN, a long short-term memory (LSTM) RNN and a gated recurrent units (GRU) RNN.

RNN prediction training steps

Setup

Requirements (tested versions)

matplotlib (2.2.2 to 3.1.1)
numpy (1.14.2 to 1.17.2)
pandas (0.22.0)
python (3.5.2 to 3.7.4)
scikit (0.22.2.post1)
torch (1.1.0)
Can be executed in CPU mode