Table 6 Performance comparison of nonlinear machine learning models using an 80-20 train-test split.
From: A quantitative study of cytotoxic compounds using graph based descriptors and machine learning
Model | MAE (Train) | MAE (Test) | MSE (Train) | MSE (Test) | RMSE (Train) | RMSE (Test) | \(R^{2}\) (Train) | \(R^{2}\) (Test) |
|---|---|---|---|---|---|---|---|---|
Decision Tree | 0.00 | 28.95 | 0.00 | 1780.48 | 0.00 | 32.10 | 1.00 | 0.79 |
k-Nearest Neighbors | 19.82 | 26.31 | 873.88 | 1223.40 | 29.56 | 34.98 | 0.94 | 0.86 |
Support Vector Regressor | 63.61 | 60.42 | 13618.71 | 8457.72 | 116.70 | 91.97 | -0.01 | 0.00 |
Voting Regressor | 9.91 | 24.41 | 218.47 | 1025.91 | 14.78 | 31.76 | 0.98 | 0.87 |
Random Forest | 14.21 | 17.49 | 511.60 | 496.70 | 22.62 | 22.29 | 0.96 | 0.94 |
Gradient Boosting | 0.00 | 15.65 | 1.03 | 588.26 | 0.00 | 24.25 | 1.00 | 0.93 |
