def _build_dataframe(self):
"""
Method to store all the information given by the class
Evaluator in a DataFrame. Here we calculate another metric
'Score*Preci' which is the product between score,
raw_score and precision of a model. The ideia is
that a good model has both higher precision (contemplate more words)
and higher score (accuracy in the analogy test).
"""
all_observations = []
for name, path in zip(self.model_names, self.pickle_paths):
if self.verbose:
print("\nEvaluating the model {}".format(name))
evaluator = Evaluator(path, self.eval_path, self.encoding)
precision, raw_score, score = evaluator.get_metrics()
observation = {}
observation['Name'] = name
observation['Raw_Score'] = raw_score
observation['Score'] = score
observation['Precision'] = precision
observation['Score*Preci'] = raw_score * score * precision
all_observations.append(observation)
self.dataframe = pd.DataFrame(all_observations)
