def fit(self, X, y, eval_set=None):
X = transforming.vectorize_and_concatenate_qa(X, self.vectorizer)
if eval_set is not None:
X_val = transforming.vectorize_and_concatenate_qa(
eval_set[0], self.vectorizer, do_fit_vectorizer=False)
eval_set = (X_val, eval_set[1])
self.classifier.fit(X,
y,
eval_set=eval_set,
early_stopping_rounds=800,
verbose=100)
def predict_proba(self, X):
if type(X) is not pd.DataFrame:
X = pd.DataFrame(np.reshape(X, (-1, 2)),
columns=['question', 'text'])
X = transforming.vectorize_and_concatenate_qa(
X, self.vectorizer, do_fit_vectorizer=False).tocsr()
return self.classifier.predict_proba(X)
def grid_search(self, X, y, param_grid, scoring=None, n_jobs=-2):
X = transforming.vectorize_and_concatenate_qa(X, self.vectorizer)
gs = GridSearchCV(estimator=self.classifier,
param_grid=param_grid,
scoring=scoring,
cv=3,
verbose=1,
n_jobs=n_jobs)
gs.fit(X, y)
return gs.best_params_
def fit(self, X, y):
X = transforming.vectorize_and_concatenate_qa(X, self.vectorizer)
self.classifier.fit(X, y)
def fit(self, X, y):
X = transforming.vectorize_and_concatenate_qa(X,
self.vectorizer).tocsr()
self.classifier.fit(X, y, verbose=200)
