jena climate forcast with models(Rnn, Cnn, Dense)

The goal in this project is to find which model is suitable for forecast in "Jena climate".
The models have Rnn(Lstm), Cnn, Dense

Find the best width for training model

1. take probable width 5~50
2. train ecah model with width 5~50
3. find the max r2_score (the best performance)
4. take correspond width to train model

Use predict data to predict future

1. predict a new data from trainned model.
2. append prediction data to our trainning dataset
3. take last same length
4. back to 1

Result - Compute R2 value of training and test

1. Training
   Rnn : 0.966580459961591
   Cnn : 0.970434208330225
   Dense : 0.975481037086605
   

2. Test
   Rnn : 0.6429095118313762
   Cnn : 0.7851424365003093
   Dense : 0.682133078239662