Deep Forest: Predicting From Averaged Forests
Intermediate
The problem, from first principles
The last Deep Forest layer makes its prediction by averaging the class vectors of all its forests, then choosing the class with the largest average. This combines several forests' opinions into one answer.
From theory to code
Implement cascade_predict(forest_probas), which averages the forests' class vectors per sample and returns the argmax class.
Constraints
- Ties are broken by the lowest class index.
Hints
Hint 1
Stack the list into one array, average over the forest axis, then take argmax along the class axis.
Keywords
research-papersclassical-mlensemblesdeep-forest
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.