Wrapper Algorithm for All Relevant Feature Selection
An all relevant feature selection wrapper algorithm.
It finds relevant features by comparing original attributes' importance with importance achievable at random, estimated using their permuted copies (shadows).