I am a bit confused about how to justify the significance of features whose weights are determined by ReliefF method in binary classification problem. Is there any certain value of weight that make some features more significant than the others?
I am working with Matlab and this link shows the example of feature's weight. Can I generalized a certain interval of value (e.g. more than 0.5) to justify that the features important on classification or should I conduct more analysis to find which feature's with their weight which are the most important to distinguish two category of data?
