Table 5.
Cross table of PCR results and predictions.
| Prediction on RNN | PCR-result | |
|---|---|---|
| Negative | Positive | |
| A. Validation of whole data | ||
| Negative | 1481 | 242 |
| Positive | 197 | 312 |
| B. Validation of undersampling data | ||
| Negative | 380 | 174 |
| Positive | 138 | 416 |
| C. Cross table on the test data predicted by whole data-learned RNN | ||
| Negative | 1419 | 73 |
| Positive | 176 | 6 |
| D. Cross table on the test data predicted by whole data-learned RNN | ||
| Negative | 1051 | 42 |
| Positive | 544 | 37 |
Cross tables on PCR-results and RNN-predictions are made on the RNN-predictions on 200-epochs. Set numbers on RNN-prediction of the test data (C, D) were shown on each learned group on which prediction algorithm was created by learning. Data from prediction on group 1 is shown on C and D, average and standard deviations on prediction from 5 groups are listed in Table 6.