def init_weight_kendall(self, feature_name, samples, order):
iw = {}
for f in feature_name:
print "Feature: %s" % f
k = evaluate_kendall(sorted(samples.keys(), key = lambda m: samples[m].feature[f], reverse = True), order)
iw[f] = k
print "Init weight by Kendall: %s" % iw
return iw
def init_weight_kendall(self, feature_name, samples, order):
iw = {}
for f in feature_name:
print "Feature: %s" % f
k = evaluate_kendall(
sorted(samples.keys(),
key=lambda m: samples[m].feature[f],
reverse=True), order)
iw[f] = k
print "Init weight by Kendall: %s" % iw
return iw
def evaluate(self):
ranked = sorted(self.samples.values(),
key=lambda m: self._weight_feature(m),
reverse=True)
ret = evaluate_kendall([m.msg_id for m in ranked], self.order)
return ret
def evaluate(self):
ranked = sorted(self.samples.values(), key = lambda m: self._weight_feature(m), reverse = True)
ret = evaluate_kendall([m.msg_id for m in ranked], self.order)
return ret
