Table 4.
Yield prediction performance of the CNN-RNN model on the subset of features. The unit of RMSE is bushels per acre.
| Response | Model | Training RMSE | Training Correlation Coefficient (%) | Validation RMSE | Validation Correlation Coefficient (%) |
|---|---|---|---|---|---|
| Corn | CNN-RNN using 50% of features | 16.86 | 89.69 | 22.24 | 80.31 |
| CNN-RNN using 75% of features | 15.0 | 91.76 | 21.65 | 79.76 | |
| CNN-RNN using 100% of features | 11.48 | 94.99 | 17.64 | 87.82 | |
| Soybean | CNN-RNN using 50% of features | 4.97 | 87.44 | 5.95 | 79.77 |
| CNN-RNN using 75% of features | 4.04 | 92.15 | 5.36 | 83.67 | |
| CNN-RNN using 100% of features | 3.85 | 92.54 | 4.91 | 87.09 |