complement_label - category 0 is always 1

[michaelgorkow]

Hi shuyo,

could you explain why you are always setting category 0 to 1? (vec[0]= 1.0)

 def complement_label(self, label):
        if not label: return numpy.ones(len(self.labelmap))
        vec = numpy.zeros(len(self.labelmap))
        vec[0] = 1.0
        for x in label: vec[self.labelmap[x]] = 1.0
        return vec

Thank you and best regards!

[datduong]

I think he has category "zero" as the "common" label. Notice this line here in the code
labelset.insert(0, "common"). Also, do you know how we can use the already trained model to make prediction for a new unseen document ?