|
| Classification algorithm | Grid search |
| Parameter | Bounds | Best parameter arrival | Best parameter departure |
|
| Probabilistic neural network | Theta minus | From 0.1 to 0.9 by step 0.1 | Theta minus: 0.1 | Theta minus: 0.1 |
| Theta plus | From 0.1 to 0.9 by step 0.1 | Theta plus: 0.5 | Theta plus: 0.8 |
| Multilayer perceptron | Iteration | From 10 to 500 by step 10 | Iteration: 500 | Iteration: 500 |
| Hidden layers number | From 1 to 4 by step 1 | Layers: 3 | Layers: 2 |
| Neurons number in the layer | From 10 to 100 by step 10 | Neurons: 60 | Neurons: 70 |
| Decision trees | Split criterion | Gain ratio; Gini index | Split: Gini index | Split: Gini index |
| Pruning method | Off: MDL | Pruning: MDL | Pruning: MDL |
| Random forest | Split criterion | Gain/gain ratio; Gini index | Split: Gini index | Split: Gini index |
| Number of models | from 50 to 1000 by step 50 | Models: 600 | Models: 500 |
| Tree ensemble | Split criterion | Gain/gain ratio; Gini index | Split: Gini index | Split: Gini index |
| Number of models | From 50 to 1000 by step 50 | Models: 450 | Models: 150 |
| Attribute sampling (columns) | No sampling; square root; linear fraction; absolute value | Sampling: absolute | Sampling: absolute |
| Fraction to learn single model | ON/OFF | Value: 9 | value: 5 |
| Fraction to learn single model | From 0.1 to 1 by step 0.1 | Fraction: OFF | Fraction: OFF |
| Gradient boosted trees | Tree depth | From 1 to 15 by step 1 | Tree depth: 12 | Tree depth: 9 |
| Number of models | From 50 to 1000 by step 50 | Models: 300 | Models: 300 |
| Learning rate | From 0.1 to 1 by step 0.1 | Learning rate: 0.2 | Learning rate: 0.2 |
| Attribute sampling (columns) | No sampling; square root; linear fraction; absolute value | Sampling: absolute Value: 10 | Sampling: absolute Value: 7 |
| Fraction to learn single model | ON/OFF | Fraction: ON | Fraction: ON |
| Fraction to learn single model | From 0.1 to 1 by step 0.1 | Fraction: 1 | Fraction: 1 |
| Support vector machines | Overlapping penalty | From 1 to 10 by step 1 | Overlapping: 3 Kernel RBF: 0.2 | Overlapping: 2 Kernel RBF: 0.4 |
| Kernel | Polynomial | Power | From 1 to 10 by step 0.5 |
| Bias | From 1 to 10 by step 0.5 |
| Gamma | From 1 to 10 by step 0.5 |
| Hypertangent | Kappa | From 0.1 to 1 by step 0.1 |
| Delta | From 0.1 to 1 by step 0.1 |
| RBF | Sigma | From 0.1 to 1 by step 0.1 |
|
