Open Access
Table 1.
Accuracy and average error obtained for each modeling approach in the construction of total tannin and anthocyanin machine learning models.
| Model | Average Error / % Accuracy | |||
|---|---|---|---|---|
| Training tannins | Test tannins | Training anthocyanins | Test anthocyanins | |
| Kernel ridge | 178.6 / 37.9% | 277.6 / 46.6% | 65.3 / 86.1% | 129.8 / 76.2% |
| Support vector machine | 87.1 / 78.8% | 277.6 / 45% | 24 / 95.7% | 135.5 / 75.5% |
| Random forest | 75.4 / 74.2 | 288.9 / 41.5% | 35.5 / 92.1% | 109.2 / 79.3% |
| Gradient boosting | 0 / 99.9% | 290.3 / 44.3% | 0 / 99.9% | 109.1 / 80.9% |
| Stochastic gradient descent | 217.4 / 24.1% | 276.5 / 42.6% | 116.8 / 74% | 116.6 / 76.6% |
| K-nearest neighbors | 0 / 100% | 288 / 42.3% | 0 / 100% | 115.1 / 77.6% |
| Gaussian processes (RBF) | 0 / 99.9% | 435.8 / 35.3% | 0 / 99.9% | 211.8 / 62% |
| Gaussian processes (Matern) | 0 / 99.9% | 335.6 / 45.7% | 0 / 99.9% | 148.9 / 73.3% |
| Gaussian processes (ExpSineSquared) | 0 / 99.9% | 404.6 / 36.4% | 0 / 99.9% | 197.2 / 65% |
| Gaussian processes (Rational quadratic) | 2.84 / 99.9% | 280 / 43.07% | 1.49 / 99.9% | 115.9 / 77.4% |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.