class Scorer(object):
classification_max = 4
cv_folds = 2
def __init__(self, text, scores):
self.text = text
self.scores = scores
self.feature_generator = FeatureGenerator()
self.classifier = RandomForestRegressor(
n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1
)
unique_scores = set(scores)
if len(unique_scores) <= self.classification_max:
self.classifier = RandomForestClassifier(
n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1
)
self.fit_feats()
self.fit_done = False
def fit_feats(self):
self.feature_generator.fit(self.text, self.scores)
def get_features(self):
feats = []
for t in self.text:
feats.append(self.feature_generator.get_features(t))
feat_mat = np.vstack(feats)
return feat_mat
def train(self):
feats = self.get_features()
scores = np.array(self.scores)
# Compute error metrics for the estimator.
self.cv_scores = cross_validation.cross_val_score(self.classifier, feats, scores)
self.cv_score = self.cv_scores.mean()
self.cv_dev = self.cv_scores.std()
self.classifier.fit(feats, scores)
self.fit_done = True
def predict(self, text):
feats = self.feature_generator.get_features(text)
return self.classifier.predict(feats)
class Scorer(object):
classification_max = 4
cv_folds = 2
def __init__(self, text, scores):
self.text = text
self.scores = scores
self.feature_generator = FeatureGenerator()
self.classifier = RandomForestRegressor(n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1)
unique_scores = set(scores)
if len(unique_scores) <= self.classification_max:
self.classifier = RandomForestClassifier(n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1)
self.fit_feats()
self.fit_done = False
def fit_feats(self):
self.feature_generator.fit(self.text, self.scores)
def get_features(self):
feats = []
for t in self.text:
feats.append(self.feature_generator.get_features(t))
feat_mat = np.vstack(feats)
return feat_mat
def train(self):
feats = self.get_features()
scores = np.array(self.scores)
# Compute error metrics for the estimator.
self.cv_scores = cross_validation.cross_val_score(
self.classifier, feats, scores)
self.cv_score = self.cv_scores.mean()
self.cv_dev = self.cv_scores.std()
self.classifier.fit(feats, scores)
self.fit_done = True
def predict(self, text):
feats = self.feature_generator.get_features(text)
return self.classifier.predict(feats)
def __init__(self, text, scores):
self.text = text
self.scores = scores
self.feature_generator = FeatureGenerator()
self.classifier = RandomForestRegressor(n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1)
unique_scores = set(scores)
if len(unique_scores) <= self.classification_max:
self.classifier = RandomForestClassifier(n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1)
self.fit_feats()
self.fit_done = False
def __init__(self, text, scores):
self.text = text
self.scores = scores
self.feature_generator = FeatureGenerator()
self.classifier = RandomForestRegressor(
n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1
)
unique_scores = set(scores)
if len(unique_scores) <= self.classification_max:
self.classifier = RandomForestClassifier(
n_estimators=100,
min_samples_split=4,
min_samples_leaf=3,
random_state=1
)
self.fit_feats()
self.fit_done = False
