Table C.1
Comparison of our CNN model with other models.
| Model | Teff | log ɡ | [Fe/H] | MG | (BP − RP)0 | Weight mean MSE |
|---|---|---|---|---|---|---|
| CNN | 7390 | 0.00497 | 0.00318 | 0.0649 | 0.00117 | 0.0685 |
| LightGBMXT | 7653 | 0.00409 | 0.00236 | 0.0531 | 0.0011 | 0.0606 |
| LightGBM | 7803 | 0.00409 | 0.00259 | 0.0558 | 0.00108 | 0.0619 |
| LightGBMLarge | 7815 | 0.0045 | 0.0033 | 0.0539 | 0.00109 | 0.0648 |
| CatBoost | 7925 | 0.00401 | 0.0023 | 0.0536 | 0.00117 | 0.0618 |
| NeuralNetFastAI | 8671 | 0.00438 | 0.00165 | 0.0615 | 0.00136 | 0.0668 |
| NeuralNetTorch | 8313 | 0.00561 | 0.00323 | 0.0672 | 0.00158 | 0.0782 |
| ExtraTreesMSE | 8470 | 0.00628 | 0.0051 | 0.0607 | 0.00128 | 0.0803 |
| RandomForestMSE | 8523 | 0.00586 | 0.00568 | 0.0614 | 0.00123 | 0.0804 |
| KNeighborsUnif | 12530 | 0.01385 | 0.02056 | 0.1644 | 0.00212 | 0.1891 |
| KNeighborsDist | 12607 | 0.01377 | 0.02015 | 0.1635 | 0.00211 | 0.1875 |
Notes. Other models are automatically tuned by AutoGluon using the same training set. The performance values are evaluated using the test set. The first five columns show the MSE values of the five parameters under different models, demonstrating the parameter prediction ability for each parameter. The last column shows the weighted mean MSE of all five parameters, reflecting the overall performance of each model.
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.