What is that enumerated loop in infer_country supposed to do?

[andreas-wolf]

Sorry, I really do not understand this code:

mordecai/mordecai/geoparse.py

Lines 716 to 738 in 05fa31d

	for loc in proced:
	feat = self.make_country_matrix(loc)
	#print("feat:", feat)
	#labels = loc['labels']
	feat_list.append(feat)
	#try:
	# for each potential country...
	for n, i in enumerate(feat_list):
	labels = i['labels']
	try:
	prediction = self.country_model.predict(i['matrix']).transpose()[0]
	ranks = prediction.argsort()[::-1]
	labels = np.asarray(labels)[ranks]
	prediction = prediction[ranks]
	except ValueError:
	print(traceback.print_exc())
	prediction = np.array([0])
	labels = np.array([""])
	loc['country_predicted'] = labels[0]
	loc['country_conf'] = prediction[0]
	loc['all_countries'] = labels
	loc['all_confidence'] = prediction

You're looping over a collection proced
For every entry you create a country matrix. Ok.
Which you append to a list. Why?
Then you loop over that list, FOR EVERY loc in proced, but you only use the last iteration of that loop to write it back to loc
Furthermore when spacy found a location which is not in geonames then i['matrix'] is an empty list and self.country_model.predict(i['matrix']) throws a ValueError. It might be wiser to test feat if it contains values before calling predict.