Entropy, in bits, of the class distribution generated by the model's predictions. Calculated by taking the sum of -log2(predictedProb) over all instances, where predictedProb is the probability (according to the model) of the actual class for that instance. If instances are weighted, the weighted sum is taken.
| Minimum value | 0 |
| Maximum value | Inf |
| Unit | bits |
| Optimization | Higher is better |