def test_unigram_all():
pos_origin_file = '../data/rule_word_origin_pos.txt'
neg_origin_file = '../data/rule_word_origin_neg.txt'
pos_stem_file = '../data/rule_word_stem_pos.txt'
neg_stem_file = '../data/rule_word_stem_neg.txt'
svm_origin_score = test_unigram(pos_origin_file, neg_origin_file, NuSVC())
svm_stem_score = test_unigram(pos_stem_file, neg_stem_file, NuSVC())
bnb_origin_score = test_unigram(pos_origin_file, neg_origin_file,
BernoulliNB())
bnb_stem_score = test_unigram(pos_stem_file, neg_stem_file, BernoulliNB())
rfc_origin_score = test_unigram(pos_origin_file, neg_origin_file,
RandomForestClassifier())
rfc_stem_score = test_unigram(pos_stem_file, neg_stem_file,
RandomForestClassifier())
if not os.path.exists('../result'):
os.mkdir('../result')
with open('../result/unigram_result.txt', 'wt', encoding='utf-8') as f:
f.write('original word result:\n')
f.write('\t\t SVM: {0:.3f}%\n'.format(svm_origin_score * 100))
f.write('\t\t BNB: {0:.3f}%\n'.format(bnb_origin_score * 100))
f.write('\t\t RFC: {0:.3f}%\n'.format(rfc_origin_score * 100))
f.write('\n stem result:\n')
f.write('\t\t SVM: {0:.3f}%\n'.format(svm_stem_score * 100))
f.write('\t\t BNB: {0:.3f}%\n'.format(bnb_stem_score * 100))
f.write('\t\t RFC: {0:.3f}%\n'.format(rfc_stem_score * 100))
def data_feature_importance(self,
features_list,
title="Feature Importance"):
from sklearn.preprocessing import LabelEncoder, Imputer
from sklearn.cross_validation import train_test_split
# extraemos las columnas con los features
clf_data = self.dataframe.loc[:, features_list]
# Preprocesaos los datos y los ajustamos
cat_feats_to_use = list(clf_data.select_dtypes(include=object).columns)
for feat in cat_feats_to_use:
encoder = LabelEncoder()
clf_data[feat] = encoder.fit_transform(clf_data[feat])
# Llenamos los valores vacios
num_feats_to_use = list(clf_data.select_dtypes(exclude=object).columns)
for feat in num_feats_to_use:
imputer = Imputer(strategy='median')
clf_data[feat] = imputer.fit_transform(
clf_data[feat].values.reshape(-1, 1))
# Separamos el index de loas Fetures
X = clf_data.iloc[:, 1:]
y = clf_data.iloc[:, 0] # the target were the first column I included
# Entrenamos con los datos recivido
x_train, _, y_train, y_test = train_test_split(X,
y,
test_size=.3,
random_state=35)
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
x_train = scaler.fit_transform(x_train)
from sklearn.ensemble.forest import RandomForestClassifier
# inicializamos el clasificador
clf = RandomForestClassifier(n_estimators=8, random_state=34)
clf.fit(x_train, y_train)
# Pasamos los datos a un Dataframe para poder graficarlos
feats_imp = pd.DataFrame(clf.feature_importances_,
index=X.columns,
columns=['FeatureImportance'])
feats_imp = feats_imp.sort_values('FeatureImportance', ascending=False)
feats_imp.plot(kind='barh', figsize=(12, 6), legend=False)
plt.title(title)
sns.despine(left=True, bottom=True)
plt.gca().invert_yaxis()
plt.savefig(self.DefeaultPath + " feature importance.png", dpi=200)
plt.cla()
plt.clf()
return
def plot_tree(profile, group, avg_acc, n_tree, picutre):
'''
选择最优的随机种子并建模
:param profile: 丰度表
:param group: 分组表
:param avg_acc: 随机种子准确率的输出文件
:param n_tree: 模型中树的颗数
:param picutre: 输出图形的名称
:return: None
'''
acc = pd.read_csv(avg_acc, sep='\t', header=0, index_col=0)
best_state = int(acc.sort_values('avgAcc').index[-1])
# 训练和保存预测模型
rf = RandomForestClassifier(n_estimators=n_tree, max_leaf_nodes=3,
random_state=best_state)
rf.fit(profile, group['label'])
joblib.dump(rf, 'rf.pkl')
# 绘制分类树结果图形
dot = picutre.split('.')[0] + '.dot'
tree_in_forest = rf.estimators_[rf.n_estimators-1]
export_graphviz(tree_in_forest,
out_file=dot,
feature_names=profile.columns,
filled=True,
rounded=False,
precision=100)
os.system('dot -Tpng {0} -o {1}'.format(dot, picutre))
def estimate_weights_random_forests(X_s, X_t, X_w):
X_all, all_labels = prepare_data_for_weights_estimation(X_s, X_t)
# train logistic regression
kf = KFold(X_all.shape[0], 10, shuffle=True)
param_grid_rf = [{
"n_estimators": np.array([500]),
"max_depth": np.array([6]),
# "max_features": np.array([1, 2, 4, 8, 16]),
"min_samples_leaf": np.array([100])
}]
rf = GridSearchCV(RandomForestClassifier(50,
max_depth=10,
class_weight="auto",
n_jobs=-1),
param_grid_rf,
cv=kf,
n_jobs=-1)
rf = RandomForestClassifier(100,
max_depth=6,
min_samples_leaf=200,
class_weight="auto",
n_jobs=-1)
rf.fit(X_all, all_labels)
# print "best parameters for rf weights determination: ", rf.best_estimator_
probas = rf.predict_proba(X_w)
weights = probas[:, 1] / probas[:, 0]
return weights
def mymap(data, N):
data = cPickle.loads(str(data))
x = data[:, :-1]
y = data[:, -1]
model = RandomForestClassifier(n_estimators=N, max_depth=6)
model = model.fit(x, y)
return cPickle.dumps(model)
def test_imdb_padded_valid(self):
num_samples = 32
num_words = 1024
(x_train, y_train), (x_test, y_test) = TestUtil.get_imdb(word_dictionary_size=num_words,
num_subsamples=num_samples)
explained_model = RandomForestClassifier(n_estimators=64, max_depth=5, random_state=1)
counter = CountVectoriser(num_words)
tfidf_transformer = TfidfTransformer()
explained_model = Pipeline([('counts', counter),
('tfidf', tfidf_transformer),
('model', explained_model)])
explained_model.fit(x_train, y_train)
model_builder = RNNModelBuilder(embedding_size=num_words, with_embedding=True,
num_layers=2, num_units=32, activation="relu", p_dropout=0.2, verbose=0,
batch_size=32, learning_rate=0.001, num_epochs=2, early_stopping_patience=128)
masking_operation = WordDropMasking()
loss = binary_crossentropy
explainer = CXPlain(explained_model, model_builder, masking_operation, loss)
x_train = pad_sequences(x_train, padding="post", truncating="post", dtype=int)
x_test = pad_sequences(x_test, padding="post", truncating="post", dtype=int, maxlen=x_train.shape[1])
explainer.fit(x_train, y_train)
eval_score = explainer.score(x_test, y_test)
train_score = explainer.get_last_fit_score()
median = explainer.predict(x_test)
self.assertTrue(median.shape == x_test.shape)
def __init__(self, initial_forrest_factor=5, n_estimators=10, **kwargs):
self._initial_forrest_size = n_estimators * initial_forrest_factor
self._final_forrest_size = n_estimators
rf_fit_args = copy(kwargs)
rf_fit_args.update({'n_estimators': self._initial_forrest_size})
self._rf = RandomForestClassifier(**rf_fit_args)
class MyRfClassifier(BaseClassifier):
def __init__(self, n_estimators, max_depth, min_samples_leaf):
self.classifier = RandomForestClassifier(
**{
'verbose': 1,
'n_estimators': n_estimators,
'max_depth': max_depth,
'min_samples_leaf': min_samples_leaf,
'n_jobs': 40
})
self.name = "rf_n{n}_md{md}_ms{ms}".format(**{
"n": n_estimators,
"md": max_depth,
"ms": min_samples_leaf
})
def get_name(self):
return self.name
def fit(self, X, y, X_t, y_t):
return self.classifier.fit(X, y)
def predict_proba(self, X):
return self.classifier.predict_proba(X)
def get_feature_importances(self, feat_names):
ipts = dict(zip(feat_names, self.classifier.feature_importances_))
return ipts
class LexicaseForestClassifier(BaseEstimator, ClassifierMixin):
def __init__(self, initial_forrest_factor=5, n_estimators=10, **kwargs):
self._initial_forrest_size = n_estimators * initial_forrest_factor
self._final_forrest_size = n_estimators
rf_fit_args = copy(kwargs)
rf_fit_args.update({'n_estimators': self._initial_forrest_size})
self._rf = RandomForestClassifier(**rf_fit_args)
def fit(self, X, y):
self._rf.fit(X, y)
for t in self._rf.estimators_:
tree_y_pred = t.predict(X)
t._error_vector = squared_error_vector(y, tree_y_pred)
final_estimators = []
for i in range(self._final_forrest_size):
final_estimators.append(epsilon_lexicase_selection(self._rf.estimators_))
self._rf.estimators_ = final_estimators
self._rf.n_estimators = self._final_forrest_size
# TODO: Set other self._rf parameters to match correct size so that predict works.
def predict(self, X, y=None):
return self._rf.predict(X)
def enemy_detection_clf():
chars = np.array(['warrior', 'warlock', 'mage', 'druid', 'rogue', 'shaman', 'paladin', 'priest', 'hunter'])
data = []
target = []
for c in chars:
p = path('images/character/new/black')
for f in os.listdir(p+'/'+c):
img = Image.open(p+'/'+c+'/'+f)
w, h = img.size
pixel = img.load()
tmp = []
for y in range(h):
for x in range(w):
tmp.append(np.float(pixel[x,y] / 255))
target.append(np.str(c))
data.append(np.array(tmp))
data = np.array(data)
#image = data.view()
#image.shape = (-1, 22, 30)
#clf = svm.SVC(gamma = 0.001)
clf = RandomForestClassifier()
clf.fit(data, target)
return clf
def Random_Forest(x_train, Y_train,n_estimators=10, criterion='gini', max_depth=None, min_samples_split=2,
min_samples_leaf=1, max_features='auto', bootstrap=True, oob_score=False,
random_state=0, min_density=None):
clf = RandomForestClassifier()
clf.fit(x_train,Y_train)
return clf
def Random_Forest(x_train, Y_train,n_estimators=10, criterion='gini', max_depth=None, min_samples_split=2,
min_samples_leaf=1, max_features='auto', bootstrap=True, oob_score=False, n_jobs=1,
random_state=None, verbose=0, min_density=None, compute_importances=None, *args):
clf = RandomForestClassifier()
clf.fit(x_train,Y_train)
return clf
def training(processed_train_csv_file):
processed_train_samples = pd.read_csv(processed_train_csv_file)
processed_train_samples = processed_train_samples.replace([np.inf, -np.inf], np.nan)
processed_train_samples = processed_train_samples.fillna(value=0)
processed_train_samples_index_lst = processed_train_samples.index.tolist()
random.shuffle(processed_train_samples_index_lst)
shuffled_train_samples = processed_train_samples.ix[processed_train_samples_index_lst]
col_names = shuffled_train_samples.columns.tolist()
col_names.remove("booking_bool")
features = shuffled_train_samples[col_names].values
labels = shuffled_train_samples["booking_bool"].values
print "Training Random Forest Classifier"
rf_classifier = RandomForestClassifier(n_estimators=150, verbose=2, learning_rate=0.1, min_samples_split=10)
rf_classifier.fit(features, labels)
print "Saving the Random Forest Classifier"
data_io.save_model(rf_classifier, model_name="rf_classifier.pkl")
print "Training Gradient Boosting Classifier"
gb_classifier = GradientBoostingClassifier(n_estimators=150, verbose=2, learning_rate=0.1, min_samples_split=10)
gb_classifier.fit(features, labels)
print "Saving the Gradient Boosting Classifier"
data_io.save_model(gb_classifier, model_name="gb_classifier.pkl")
print "Training SGD Classifier"
sgd_classifier = SGDClassifier(loss="modifier_huber", verbose=2, n_jobs=-1)
sgd_classifier.fit(features, labels)
print "Saving the SGD Classifier"
data_io.save_model(sgd_classifier, model_name="sgd_classifier.pkl")
def test_nlp_erroneous_rnn_args_invalid(self):
num_words = 1024
(x_train,
y_train), (x_test,
y_test) = TestUtil.get_random_variable_length_dataset(
max_value=num_words)
explained_model = RandomForestClassifier(n_estimators=64,
max_depth=5,
random_state=1)
counter = CountVectoriser(num_words)
tfidf_transformer = TfidfTransformer()
explained_model = Pipeline([('counts', counter),
('tfidf', tfidf_transformer),
('model', explained_model)])
explained_model.fit(x_train, y_train)
with self.assertRaises(ValueError):
_ = RNNModelBuilder(
with_embedding=True,
verbose=0) # Must also specify the embedding_size argument.
model_builder = RNNModelBuilder(embedding_size=num_words,
with_embedding=True,
verbose=0)
input_layer = Input(shape=(10, 2))
with self.assertRaises(ValueError):
model_builder.build(input_layer)
input_layer = Input(shape=(10, 3))
with self.assertRaises(ValueError):
model_builder.build(input_layer)
def __init__(self, path):
'''
Constructor
'''
self.path = path
self.model = RandomForestClassifier(n_estimators=150,n_jobs=8)
self.model_name = 'rf'
def test_RandomForest(self):
X = [[0, 1], [1, 1]]
Y = [0, 1]
regression = RandomForestClassifier(n_estimators=10)
regression = regression.fit(X, Y)
regression.predict_proba(X)
def importance():
df = pd.read_csv(path.NOTEBOOKS_DATA + 'features3.csv')
features = list(df.columns.values)
target = 'm_column_result'
features.remove('m_match_id')
features.remove('m_column_result')
features.remove('m_match_date')
features.remove('m_goals_home')
features.remove('m_goals_away')
features.remove('a_next_match_id')
features.remove('h_next_match_id')
features.remove('m_favorite')
features.remove('m_medium')
features.remove('m_underdog')
features.remove('h_last_match_local')
features.remove('a_last_match_local')
features.remove('rf1000')
features.remove('rf1000_fs1')
X = df[features]
y = df[target]
# fit an Extra Trees model to the data
clf = RandomForestClassifier(n_estimators=1000)
clf.fit(X, y)
# display the relative importance of each attribute
for x,y in zip(features,clf.feature_importances_):
print (x,y)
def stkFoldCrossValidation():
X = pickle.load(open('X.p', 'rb'))
X = np.array(X)
Y = pickle.load(open('Y.p', 'rb'))
Y = np.array(Y)
skf = StratifiedKFold(n_splits=10)
skf.get_n_splits(X, Y)
k = 1
for train_index, test_index in skf.split(X, Y):
X_train, X_test = X[train_index], X[test_index]
Y_train, Y_test = Y[train_index], Y[test_index]
print(k)
k += 1
rf = RandomForestClassifier()
rf.fit(X_train, Y_train)
yp = rf.predict(X_test)
print(classification_report(Y_test, yp, digits=6))
def rforest_classify(X,Y):
#clf = RandomForestClassifier(criterion='gini',max_features=7,n_estimators=100,n_jobs=3,min_samples_leaf=5)
clf = RandomForestClassifier(n_estimators=500, \
criterion='entropy', max_depth=None, min_samples_split=2, min_samples_leaf=1 \
,max_features='auto', bootstrap=False, oob_score=False, n_jobs=-1, min_density=None)
clf.fit(X,Y)
return clf
def RF_Features_Importance(X, Y, outputfile="RF.csv"):
forest = RandomForestClassifier(n_estimators=300)
forest.fit(X, Y)
importances = np.matrix(forest.feature_importances_).tolist()[0]
df = pd.DataFrame(list(zip(header, importances)),
columns=["Features", "Importance"])
df.to_csv(outputfile, index=False)
def forest(X, y, model_path):
model = RandomForestClassifier()
model.fit(X, y)
expected = y
predicted = model.predict(X)
print(metrics.classification_report(expected, predicted))
print(metrics.confusion_matrix(expected, predicted))
joblib.dump(model, model_path)
def __init__(self, sig_weight=1., pow_sig=1., pow_bg=1., n_estimators=10,
criterion="gini", max_depth=None, min_samples_split=2, min_samples_leaf=1, max_features="auto",
bootstrap=True, oob_score=False, n_jobs=1, random_state=None, verbose=0, min_density=None,
compute_importances=None):
RandomForestClassifier.__init__(self)
# Everything should be set via set_params
self.sig_weight = sig_weight
self.pow_bg = pow_bg
self.pow_sig = pow_sig
def calcRandomForestClassifier(channels_training, channels_testing,
target_training, target_testing):
clf = RandomForestClassifier(n_estimators=500,
max_features=int(
sqrt(len(channels_training[0]))))
clf = clf.fit(channels_training, target_training)
predictions = clf.predict(channels_testing)
comp = [predictions, target_testing, channels_testing]
return clf, comp
def calc_score(test, train):
test_f, test_l = split_data_label(test)
train_f, train_l = split_data_label(train)
# 학습시키고 정답률 구하기
clf = RandomForestClassifier()
clf.fit(train_f, train_l)
pre = clf.predict(test_f)
return metrics.accuracy_score(test_l, pre)
def train_rf(train_vec, train_label):
from sklearn.ensemble.forest import RandomForestClassifier as RFC
# rfrclf = RFR(n_estimators=1001)
# rfrclf.fit(train_vec, train_label)
# print rfrclf.feature_importances_
trfclf = RFC(n_estimators=1001)
trfclf.fit(train_vec, train_label)
# print rfclf.feature_importances_
return trfclf
def drawfeature(train_data_path,train_file_name,test_data_path,test_file_name):
train_file = os.path.join(train_data_path,train_file_name)
train_data = pd.read_csv(train_file)
n_data_train = train_data['text'].size
print 'n_data_train is %s' %n_data_train
print type(n_data_train)
test_file = os.path.join(test_data_path,test_file_name)
test_data = pd.read_csv(test_file)
n_data_test = test_data['text'].size
print 'n_data_test is %s' %n_data_test
print type(n_data_test)
vectorizer = CountVectorizer(analyzer='word',tokenizer = None,
preprocessor = None, stop_words=None, max_features = 5000)
transformer = TfidfTransformer()
train_data_words = []
print 'start with words in train data set'
for i in xrange(n_data_train):
if((i+1)%1000 == 0):
print 'Drawfeatures line %d of %d' %(i+1,n_data_train)
train_data_words.append(words_to_features(train_data['text'][i]))
print 'start bag of words in train data....'
train_data_features = vectorizer.fit_transform(train_data_words)
train_data_features = train_data_features.toarray()
print 'start tfidf in train data....'
train_data_features = transformer.fit_transform(train_data_features)
train_data_features = train_data_features.toarray()
#test-data processing
test_data_words = []
for i in xrange(n_data_test):
if((i+1)%1000 == 0):
print 'Drawfeatures line %d of %d' %(i+1,n_data_test)
test_data_words.append(words_to_features(test_data['text'][i]))
test_data_features = vectorizer.fit_transform(test_data_words)
test_data_features = test_data_features.toarray()
print'randome forest go...'
forest = RandomForestClassifier(n_estimators = 13)
forest = forest.fit(train_data_features,train_data['label'])
pred = forest.predict(test_data_features)
pred = pd.Series(pred,name='Target')
pred.to_csv('SENTI_RF.CSV',index=None, header = None)
print'naive baby go...'
mnb = MultinomialNB(alpha=0.01)
mnb = mnb.fit(train_data_features,train_data['label'])
pred = mnb.predict(test_data_features)
pred = pd.Series(pred,name = 'Target')
pred.to_csv('SENTI_MNB',index = None, header = True)
def main(args):
if args.analyse != None:
train_data_x, test_data_x,train_data_y, test_data_y = process_data(args.analyse)
RT = RandomForestClassifier(n_estimators=100)
RT.fit(train_data_x, train_data_y)
print RT.score(test_data_x, test_data_y)
return
def random_forest_classifier(features, target):
"""
To train the random forest classifier with features and target data
:param features:
:param target:
:return: trained random forest classifier
"""
clf = RandomForestClassifier(n_estimators=600, max_depth=50)
clf.fit(features, target)
return clf
def my_digits():
digits = _data()
n_samples = len(digits.images)
datas = digits.images.reshape((n_samples, -1))
classifier = RandomForestClassifier()
classifier.fit(datas, digits.target)
return classifier
def test_unigram_all():
# test_unigram(NuSVC(), 'SVM', '../data/pos_origin.txt', '../data/neg_origin.txt', 'origin')
# test_unigram(NuSVC(), 'SVM', '../data/pos_stem.txt', '../data/neg_stem.txt', 'stem')
test_unigram(BernoulliNB(), 'BNB', '../data/pos_origin.txt',
'../data/neg_origin.txt', 'origin')
test_unigram(BernoulliNB(), 'BNB', '../data/pos_stem.txt',
'../data/neg_stem.txt', 'stem')
test_unigram(RandomForestClassifier(), 'RFC', '../data/pos_origin.txt',
'../data/neg_origin.txt', 'origin')
test_unigram(RandomForestClassifier(), 'RFC', '../data/pos_stem.txt',
'../data/neg_stem.txt', 'stem')
def RandomForestClassifer(self):
'''
Function to do RandomForest Classifer.
'''
train_Array = self.titanic_train_frame.values
self.test_Array = self.titanic_test_frame.values
randomForest = RandomForestClassifier(n_estimators = 100, n_jobs = -1)
randomForest.fit(train_Array[0::,1::],train_Array[0::,0])
self.predicted_probability = randomForest.predict(self.test_Array[0::,0::])
self.predicted_probability_list = self.predicted_probability.tolist()
class RFClassifier(super.abstract_classifier):
def __init__(self, train_features, train_labels, num_of_trees):
self.train_features = train_features
self.train_labels = train_labels
self.rf_member = RandomForestClassifier(num_of_trees)
def train(self):
self.rf_member.fit(self.train_features, self.train_labels)
def classify(self, newVector):
return self.rf_member.predict(newVector)
def evalOne(enabledColumns):
features = [all_features[i] for i in range(0, len(all_features)) if enabledColumns[i]]
Y = []
P = []
for group in range(0,5):
# print("Test group " + str(group + 1))
trainStationList = []
testStationList = []
for i in range(0,5):
if i == group:
testStationList.extend(groups[i])
else:
trainStationList.extend(groups[i])
trainStations = set(float(station) for station in trainStationList)
# reorder train stations
# print("\ttrainStationList:" + str(trainStationList))
trainStationList = [s for s in all_stations if float(s) in trainStations]
# print("\ttrainStationList:" + str(trainStationList))
testStations = set(float(station) for station in testStationList)
# print("\ttestStationList:" + str(testStationList))
trainX, testX, trainY, testY, trainLocation, testLocation = splitDataForXValidationWithLocation(trainStations, testStations, "location", data, features, "target")
train_lower = [float(trainStationList[i]) for i in range(0, len(trainStationList)) if i < (len(trainStationList) / 2.0)]
# train_upper = [float(trainStationList[i]) for i in range(0, len(trainStationList)) if i >= (len(trainStationList) / 2.0)]
test_lower = [float(testStationList[i]) for i in range(0, len(testStationList)) if i < (len(testStationList) / 2.0)]
# test_upper = [float(testStationList[i]) for i in range(0, len(testStationList)) if i >= (len(testStationList) / 2.0)]
trainY = []
for l in trainLocation:
if l in train_lower:
trainY.append(0)
else:
trainY.append(1)
testY = []
for l in testLocation:
if l in test_lower:
testY.append(0)
else:
testY.append(1)
model = RandomForestClassifier(random_state=42, n_estimators=20, max_depth=9, n_jobs=-1)
model.fit(trainX, trainY)
predY = model.predict(testX)
Y.extend(testY)
P.extend(predY)
f1 = f1_score(Y, P)
accuracy = accuracy_score(Y, P)
return f1, accuracy
def do_training(processed_train_csv_file):
## Processed train samples reading
# read saved processed train samples from the given csv file
processed_train_samples = pd.read_csv(processed_train_csv_file)
# inf to nan
processed_train_samples = processed_train_samples.replace([np.inf, -np.inf], np.nan)
# nan to 0
processed_train_samples = processed_train_samples.fillna(value=0)
processed_train_samples_index_lst = processed_train_samples.index.tolist()
# 之前排过序，这里shuffle一下，效果更好
random.shuffle(processed_train_samples_index_lst)
# organize new train samples and targets
shuffled_train_samples = processed_train_samples.ix[processed_train_samples_index_lst]
col_names = shuffled_train_samples.columns.tolist()
col_names.remove("booking_bool")
features = shuffled_train_samples[col_names].values
labels = shuffled_train_samples['booking_bool'].values
# Model training
# 1 Random Forest Classifier
print("Training Random Forest Classifier")
rf_classifier = RandomForestClassifier(n_estimators=150,
verbose=2,
n_jobs=-1,
min_samples_split=10)
rf_classifier.fit(features, labels)
print("Saving the Random Forest Classifier")
data_io.save_model(rf_classifier, model_name='rf_classifier.pkl')
# 2 Gradient Boosting Classifier
print("Gradient Boosting Classifier")
gb_classifier = GradientBoostingClassifier(n_estimators=150,
verbose=2,
learning_rate=0.1,
min_samples_split=10)
gb_classifier.fit(features, labels)
print("Saving the Gradient Boosting Classifier")
data_io.save_model(gb_classifier, model_name='gb_classifier.pkl')
# 3 SGD Classifier
print("SGD Classifier")
sgd_classifier = SGDClassifier(loss="modified_huber", verbose=2,
n_jobs=-1)
sgd_classifier.fit(features, labels)
print("saved the SGD Classifier")
data_io.save_model(sgd_classifier, model_name='sgd_classifier.pkl')
def train_classifier(vocal_frames, non_vocal_frames):
frames = np.append(vocal_frames, non_vocal_frames, axis=0)
labels_vocal = np.ones(vocal_frames.shape[0])
labels_non_vocal = np.zeros(non_vocal_frames.shape[0])
labels = np.append(labels_vocal, labels_non_vocal, axis=0)
rfc = RandomForestClassifier(n_estimators=100, max_depth=None)
rfc.fit(frames, labels)
return rfc
def RF(pth):
train_desc=np.load(pth+'/training_features.npy')
nbr_occurences = np.sum( (train_desc > 0) * 1, axis = 0)
idf = np.array(np.log((1.0*len(image_paths)+1) / (1.0*nbr_occurences + 1)), 'float32')
# Scaling the words
stdSlr = StandardScaler().fit(train_desc)
train_desc = stdSlr.transform(train_desc)
modelRF=RandomForestClassifier(n_estimators=10,
max_depth=5,max_features=1,random_state=0)
modelRF.fit(train_desc,np.array(train_labels))
joblib.dump((modelRF, img_classes, stdSlr), pth+"/rf-bof.pkl", compress=3)
test(pth, "rf-")
def train(self, log_probs, labels):
'''
Train the stacked classifier
@param log_probs: list of probability matrices (channels, label)
@param labels: label for each feature-vector
'''
print 'Train stacked classifier with %d windows' % labels.shape[0]
log_probs = [self.channel_sort.sort(f) for f in log_probs]
log_probs = np.vstack(log_probs)
# TODO: classifier as Parameter
self.stacked_classifier = RandomForestClassifier(n_estimators=10)
self.stacked_classifier.fit(log_probs, labels)
def train_model(X_train,y_train):
print("training the model ...")
# create sets for probability calibration
X_train_train, X_prob_cal, y_train_train, y_prob_cal = train_test_split(X_train,
y_train,
test_size=0.2)
rf = RandomForestClassifier(
max_features="auto",
n_estimators=2000,
max_depth=8,
n_jobs=-1,
class_weight = 'balanced',
verbose=1)
rf.fit(X_train_train,y_train_train)
# feature importances
# feature_importance = False
# if(feature_importance):
#
# importances = rf.feature_importances_
# std = np.std([tree.feature_importances_ for tree in rf.estimators_],
# axis=0)
# indices = np.argsort(importances)[::-1]
# col_names = df.drop('bin',axis=1).columns.values
# print("Feature ranking:")
#
# for f in range(X_train_train.shape[1]):
# print("%d. %s (%f)" % (f + 1, col_names[indices[f]], importances[indices[f]]))
#
# # Plot the feature importances of the forest
# plt.figure()
# plt.title("Feature importances")
# plt.bar(range(X_train_train.shape[1]), importances[indices],
# color="r", yerr=std[indices], align="center")
# plt.xticks(range(X_train_train.shape[1]), col_names[indices],rotation = 50)
# plt.xlim([-1, X_train_train.shape[1]])
# plt.show()
# Probability calibration
sig_clf = CalibratedClassifierCV(rf, method="sigmoid", cv="prefit")
sig_clf.fit(X_prob_cal, y_prob_cal)
y_pred_train = sig_clf.predict_proba(X_train)
print(".. training log_loss : {:0.2f} %".format(log_loss(y_train,y_pred_train)*100))
return sig_clf
class Model(BaseModel):
"""Antares implementation of scikit learn random forest classifier
"""
def __init__(self,
categorical_features=None,
n_estimators=50,
n_jobs=-1,
max_depth=10):
'''
Example:
>>> from madmex.modeling.supervised.rf import Model
>>> rf = Model()
>>> # Write model to db
>>> rf.to_db(name='test_model', recipe='mexmad', training_set='no')
>>> # Read model from db
>>> rf2 = Model.from_db('test_model')
'''
super().__init__(categorical_features=categorical_features)
self.model = RandomForestClassifier(n_estimators=n_estimators,
n_jobs=n_jobs,
max_depth=max_depth)
self.model_name = 'rf'
def fit(self, X, y):
X = self.hot_encode_training(X)
self.model.fit(X, y)
def predict(self, X):
'''
Simply passes down the prediction from the underlying model.
'''
X = self.hot_encode_predict(X)
return self.model.predict(X)
def predict_confidence(self, X):
"""Get confidence of every prediction
"""
X = self.hot_encode_predict(X)
return self.model.predict_proba(X).max(axis=1)
def score(self, X, y):
'''
Test the model given a dataset and a target vector.
This method applies the model that this object represents to the given dataset using
the response variable y. It is a measure of the accuracy of the trained model. Usually
the orginal dataset should be splitted in training and testing subsets to cross validate
the model.
'''
return self.model.score(X, y)
def classic_model(image_dir, image_lists, method):
X, y = get_X_y(image_dir, image_lists, ['training', 'validation'], method)
classifier = RandomForestClassifier(n_estimators=1000, n_jobs=4)
classifier.fit(X, y)
X_test, y_test = get_X_y(image_dir, image_lists, ['testing'], method)
predictions = classifier.predict(X_test)
confusion = pandas.crosstab(y_test,
predictions,
rownames=['Actual Class'],
colnames=['Predicted Class'])
print confusion
return accuracy_score(y_test, predictions)
def __init__(self, n_estimators, max_depth, min_samples_leaf):
self.classifier = RandomForestClassifier(
**{
'verbose': 1,
'n_estimators': n_estimators,
'max_depth': max_depth,
'min_samples_leaf': min_samples_leaf,
'n_jobs': 40
})
self.name = "rf_n{n}_md{md}_ms{ms}".format(**{
"n": n_estimators,
"md": max_depth,
"ms": min_samples_leaf
})
def try_model(train):
print(train.shape)
features = ["phone_brand", "device_model", "event_count", "action_radius_max", "medianTime", "minTime", "maxTime", "weekday", "appcounts1"]
encoder = LabelEncoder()
train["group"] = encoder.fit_transform(train["group"].values)
rf = RandomForestClassifier(n_estimators=50, max_depth=15, max_features=6, bootstrap=True, n_jobs=4, random_state=2016, class_weight=None)
rf.fit(train[features].values, train["group"].values)
feature_importance(rf, features)
skf = StratifiedKFold(train["group"].values, n_folds=5, shuffle=True, random_state=2016)
scores = cross_val_score(rf, train[features].values, train["group"].values, scoring="log_loss", cv=skf, n_jobs=1)
print(scores)
print("RF Score: %0.5f" %(-scores.mean())) # RF Score: 2.39884
def just_pred(x, y):
xlen = len(x)
i = range(xlen)
np.random.shuffle(i)
trainpct = 0.7
trainlen = int(trainpct * xlen)
testlen = xlen - trainlen
xtrain = x.ix[:trainlen,:]
ytrain = y.ix[:trainlen]
xtest = x.ix[trainlen:,:]
ytest = y.ix[trainlen:]
rf = RandomForestClassifier()
rf.fit(xtrain, ytrain)
ypred = rf.predict(xtest)
return ytest, ypred
def crossval(x, y, k=5):
for i in range(k):
i = range(len(X))
np.random.shuffle(i)
xlen = len(x)
trainpct = 0.7
trainlen = int(trainpct * xlen)
testlen = xlen - trainlen
xtrain = x.ix[:trainlen,:]
ytrain = y.ix[:trainlen]
xtest = x.ix[trainlen:,:]
ytest = y.ix[trainlen:]
rf = RandomForestClassifier()
rf.fit(xtrain, ytrain)
ypred = rf.predict(xtest)
print ypred
def __init__(self, n_estimators, max_depth, min_samples_leaf):
self.classifier = RandomForestClassifier(**{'verbose':1, 'n_estimators': n_estimators,
'max_depth':max_depth,'min_samples_leaf':min_samples_leaf,
'n_jobs':40})
self.name = "rf_n{n}_md{md}_ms{ms}".format(
**{"n": n_estimators, "md": max_depth, "ms": min_samples_leaf}
)
def __init__(self, verbose=1, n_estimators = 2000, max_depth=8, min_samples_leaf=10000,
n_jobs=25):
self.classifier = RandomForestClassifier( **{'verbose': verbose,
'n_estimators': n_estimators,
'max_depth': max_depth, 'min_samples_leaf': min_samples_leaf,
'n_jobs': n_jobs})
self.name = "rf_n{n}_md{md}_ms{ms}".format(
**{"n": n_estimators, "md": max_depth, "ms": min_samples_leaf}
)
class MyRfClassifier(BaseClassifier):
def __init__(self, n_estimators, max_depth, min_samples_leaf):
self.classifier = RandomForestClassifier(**{'verbose':1, 'n_estimators': n_estimators,
'max_depth':max_depth,'min_samples_leaf':min_samples_leaf,
'n_jobs':40})
self.name = "rf_n{n}_md{md}_ms{ms}".format(
**{"n": n_estimators, "md": max_depth, "ms": min_samples_leaf}
)
def get_name(self):
return self.name
def fit(self, X, y, X_t, y_t):
return self.classifier.fit(X, y)
def predict_proba(self, X):
return self.classifier.predict_proba(X)
def get_feature_importances(self, feat_names):
ipts = dict(zip(feat_names, self.classifier.feature_importances_))
return ipts
class MyRandomForestClassifier(BaseClassifier):
def __init__(self, verbose=1, n_estimators = 2000, max_depth=8, min_samples_leaf=10000,
n_jobs=25):
self.classifier = RandomForestClassifier( **{'verbose': verbose,
'n_estimators': n_estimators,
'max_depth': max_depth, 'min_samples_leaf': min_samples_leaf,
'n_jobs': n_jobs})
self.name = "rf_n{n}_md{md}_ms{ms}".format(
**{"n": n_estimators, "md": max_depth, "ms": min_samples_leaf}
)
def get_name(self):
return self.name
def fit(self, X, y):
return self.classifier.fit(X, y)
def predict_proba(self, X):
return self.classifier.predict_proba(X)
def get_feature_importances(self, feat_names):
return self.classifier.feature_importances_
def initDecTrees(self, path):
for filename in os.listdir(path):
if filename=='train.csv':
with open(os.path.join(path,filename)) as infile:
f = csv.reader(infile)
aux = f.next() # skip the header
x = []
y = []
for line in f:
if size(line) > 1:
if self.option == 1:
data = [converter(line[2]), converter(line[3]), converter(line[4]), converter(line[7]), converter(line[9])]
y.append(converter(line[6]))
x.append(data)
elif self.option == 2:
auxDeputy = fetchDeputyParty(line[2])
data = [converter(line[2]), converter(line[3]), converter(line[4]), converter(line[7]), converter(line[9]), encodeParty(auxDeputy['party']), encodeState(auxDeputy['state'])]
y.append(converter(line[6]))
x.append(data)
clf = RandomForestClassifier(n_estimators=5)
clf.fit(x, y)
return clf
def drawfeature(train_data_path='./train', train_filename='train_cleaned',test_data_path='./test', test_filename='test_cleaned'):
train_file = os.path.join(train_data_path, train_filename)
train_data = pd.read_csv(train_file)
n_train_data = train_data['text'].size
test_file = os.path.join(test_data_path,test_filename)
test_data = pd.read_csv(test_file)
n_test_data = test_data['text'].size
vectorizer = CountVectorizer(analyzer="word",tokenizer=None, preprocessor=None, stop_words=None, max_features=2000)
transformer = TfidfTransformer()
train_data_words = []
for i in xrange(n_train_data):
train_data_words.append(words_to_features(train_data['text'][i]))
train_data_features = vectorizer.fit_transform(train_data_words)
train_data_features = train_data_features.toarray()
train_data_features = transformer.fit_transform(train_data_features)
train_data_features = train_data_features.toarray()
train_data_pd=pd.Series(train_data_features,name=None)
train_data_pd.to_csv("trainfeature.csv", index=None, header=True)
test_data_words = []
for i in xrange(n_test_data):
test_data_words.append(words_to_features(test_data['text'][i]))
test_data_features = vectorizer.fit_transform(test_data_words)
test_data_features = test_data_features.toarray()
test_data_features = transformer.fit_transform(test_data_features)
test_data_features = test_data_features.toarray()
test_data_pd=pd.Series(test_data_features,name=None)
test_data_pd.to_csv("testfeature.csv", index=None, header=True)
forest = RandomForestClassifier(n_estimators=60)
forest = forest.fit(train_data_features, train_data['lable'])
pred = forest.pedict(test_data_features)
pred = pd.Series(pred,name='Target')
pred.to_csv("bow_tfidf_RF.csv", index=None, header=True)
audit_y = audit_y.astype(int)
print(audit_X.dtype, audit_y.dtype)
def predict_audit(classifier):
adjusted = DataFrame(classifier.predict(audit_X), columns = ["Adjusted"])
adjusted_proba = DataFrame(classifier.predict_proba(audit_X), columns = ["probability_0", "probability_1"])
return pandas.concat((adjusted, adjusted_proba), axis = 1)
audit_tree = DecisionTreeClassifier(random_state = 13, min_samples_leaf = 5)
audit_tree.fit(audit_X, audit_y)
store_pkl(audit_tree, "DecisionTreeAudit.pkl")
store_csv(predict_audit(audit_tree), "DecisionTreeAudit.csv")
audit_forest = RandomForestClassifier(random_state = 13, min_samples_leaf = 5)
audit_forest.fit(audit_X, audit_y)
store_pkl(audit_forest, "RandomForestAudit.pkl")
store_csv(predict_audit(audit_forest), "RandomForestAudit.csv")
audit_regression = LogisticRegression()
audit_regression.fit(audit_X, audit_y)
store_pkl(audit_regression, "RegressionAudit.pkl")
store_csv(predict_audit(audit_regression), "RegressionAudit.csv")
#
# Multi-class classification
#
test_size=0.25,
random_state=666)
train_ind = X_train.index.values
test_ind = X_test.index.values
g_train = g.iloc[train_ind,:]
g_test = g.iloc[test_ind,:]
clf = tree.DecisionTreeClassifier(criterion='gini',
max_depth=6,
min_samples_leaf=3)
####################
clf = RandomForestClassifier(criterion='gini',
max_depth = 6,
min_samples_leaf=3,
n_estimators=50)
####################
clf = AdaBoostClassifier(DecisionTreeClassifier(criterion='gini',
max_depth=6,
min_samples_leaf=3),
n_estimators = 200,
learning_rate = 0.1)
####################
clf = neighbors.KNeighborsClassifier(100, weights='uniform')
clf = neighbors.KNeighborsClassifier(100, weights='distance')
####################
clf = GaussianNB()
##############################
t0 = time()
param_grid = {'C': [150, 500, 750, 1000],
### the training data (features_train, labels_train) have both "fast" and "slow"
### points mixed together--separate them so we can give them different colors
### in the scatterplot and identify them visually
grade_fast = [features_train[ii][0] for ii in range(0, len(features_train)) if labels_train[ii]==0]
bumpy_fast = [features_train[ii][1] for ii in range(0, len(features_train)) if labels_train[ii]==0]
grade_slow = [features_train[ii][0] for ii in range(0, len(features_train)) if labels_train[ii]==1]
bumpy_slow = [features_train[ii][1] for ii in range(0, len(features_train)) if labels_train[ii]==1]
#### initial visualization
plt.xlim(0.0, 1.0)
plt.ylim(0.0, 1.0)
plt.scatter(bumpy_fast, grade_fast, color = "b", label="fast")
plt.scatter(grade_slow, bumpy_slow, color = "r", label="slow")
plt.legend()
plt.xlabel("bumpiness")
plt.ylabel("grade")
### your code here! name your classifier object clf if you want the
clf=RandomForestClassifier()
clf.fit(features_train,labels_train)
### draw the decision boundary with the text points overlaid
prettyPicture(clf, features_test, labels_test)
plt.show()
################################################################################
def runns(resp_var, size_of_test_data,dataset,positive_class,n_estimators,important_features,dealing_with_nulls):
dataset = pd.read_csv('raw_data.csv', low_memory=False) # For testing purposes
#----DATA PREPROCESSING
#-------dealing with NULL values in the data
#----------remove the rows in which the response is null
dataset=dataset.dropna(subset=[resp_var])
#----------dealing with nulls
dataset=deal_with_nulls(dealing_with_nulls,dataset)
#----FEATURE SELECTION
#-------get predictors important in predicting the response
#-----------transform categorical predictors to dummy variables
predictors=dataset.drop(resp_var,axis=1,inplace=False)
predictors=pd.get_dummies(predictors)
#-----------balance the classes in the response var
ros = RandomOverSampler(random_state=0)
resp=dataset[resp_var]
prds, resp = ros.fit_sample(predictors, resp)
#-----------fit the random forest classifier to give us the important predictors
rf_clf = RandomForestClassifier(n_estimators=n_estimators)
rf_clf.fit(prds,resp)
#-------get the important predictors
feature_imp = pd.Series(rf_clf.feature_importances_,
index=list(predictors.iloc[:,0:])).sort_values(ascending=False)
#-------names of the important predictors
important_predictor_names = feature_imp.index[0:important_features]
#-------subset the data to get only the important predictors and the response
resp=pd.DataFrame(data=resp,columns=[resp_var])
predictors=pd.DataFrame(prds,columns=list(predictors))
dataset=pd.concat([resp,predictors],axis=1)
#---------------------------------------------------------
#----MODEL TRAINING
#--------Remove the response variables from the features variables - axis 1 refers to the columns
m_data= dataset.drop(resp_var, axis = 1,inplace=False)
# Response variables are the values we want to predict
resp_var = np.array(dataset[resp_var])
dataset = pd.get_dummies(m_data)
# Saving feature names for later use
feature_list = list(m_data.columns)
# Convert to numpy array
dataset = np.array(dataset)
# Split the data into training and testing sets
train_features, test_features, train_labels, test_labels = train_test_split(dataset, resp_var, test_size = size_of_test_data, random_state = 402)
# Instantiate model with n_estimators decision trees
clf = SVC(kernel='rbf',probability=True)
# Train the model on training data
clf.fit(train_features, train_labels)
# evaluation
predicted = clf.predict(test_features)
pred_prob = clf.predict_proba(test_features)
accuracy = accuracy_score(test_labels, predicted)
#confusion matrix
cnf = (confusion_matrix(test_labels,predicted))
#precision score
precision = precision_score(test_labels,predicted,pos_label=positive_class)
#avg pres
avg_precision = average_precision_score(test_labels,pred_prob[:,[1]])
#recall score
rec = recall_score(test_labels,predicted,pos_label=positive_class)
#f1 scorea
fscore = f1_score(test_labels,predicted,pos_label=positive_class)
#fbeta score
fbeta = fbeta_score(test_labels,predicted,beta=0.5)
#hamming_loss
hamming = hamming_loss(test_labels,predicted)
#jaccard similarity score
jaccard = jaccard_similarity_score(test_labels,predicted)
#logloss
logloss = log_loss(test_labels,predicted)
#zero-oneloss
zero_one = zero_one_loss(test_labels,predicted)
#auc roc
area_under_roc = roc_auc_score(test_labels,pred_prob[:,[1]])
#cohen_score
cohen = cohen_kappa_score(test_labels,predicted)
#mathews corr
mathews = matthews_corrcoef(test_labels,predicted)
# Variable importances from the important features selection stage
variable_importance_list = list(zip(prds, feature_imp))
output={"accuracy":accuracy,"precision":precision,"average precision":avg_precision,"recall":rec,"fscore":fscore,"fbeta":fbeta,"hamming":hamming,"jaccard":jaccard,"logloss":logloss,"zero_one":zero_one,"area_under_roc":area_under_roc,"cohen":cohen,"mathews":mathews}
output=json.dumps(output)
return jsonify({"Predictions": output})
#'Date',
'WnvPresent_DateTrapSpecies'], axis=1)
# Create dfs based on mask
X_train = X[~msk]
X_test = X[msk]
y_train = y[~msk]
y_test = y[msk]
return X_train, X_test, y_train, y_test
# Create classifiers
clf = RandomForestClassifier(n_estimators=500,
min_samples_leaf=5)
clf = xgbwrapper.XgbWrapper({'objective': 'binary:logistic',
'eval_metric': 'auc',
'eta': 0.05,
'silent': 1})
# Cross validation
if do_cross_val:
# Leave-one-year-out cross-validation
scores = []
total_pred = np.array([])
total_test = np.array([])
def model_pred(trainX,trainY,testX,model_type):
if model_type == "rf":
clf = RandomForestClassifier(n_estimators = 500,n_jobs = 20)
clf.fit(trainX,trainY)
pred = clf.predict(testX)
if model_type == "gbdt":
clf = GradientBoostingClassifier(n_estimators=6,learning_rate=0.9,random_state=0)
clf.fit(trainX,trainY)
pred = clf.predict(testX)
if model_type == "fusion":
prob = np.zeros(len(testX))
params = [100,200,300,400,500]
for param in params:
clf = RandomForestClassifier(n_estimators = param,n_jobs = 20,bootstrap=True)
clf.fit(trainX,trainY)
prob += clf.predict(testX)
'''
params = [1,2,3,4,5,6,7,8,9,10]
for param in params:
clf = GradientBoostingClassifier(n_estimators=param,learning_rate=0.9,random_state=0)
clf.fit(trainX,trainY)
prob += clf.predict(testX)
'''
pred = list(prob >= 3)
print "the pos rate is:",float(sum(pred))/len(pred)
return pred
if 'FIGURES' in line:
break
vals = line.strip().split('\t')
text = vals[2]
corpus_test.append(text)
if int(vals[0]) == 0:
y_test.append('0')
else:
y_test.append('1')
X_train = vectorizer.fit_transform(corpus_train)
X_test = vectorizer.transform(corpus_test)
clf = RandomForestClassifier(n_estimators=10)
#clf = KNeighborsClassifier(n_neighbors=10)
#clf = LinearSVC()
clf.fit(X_train, y_train)
print len(y_train)
print len(y_test)
pred = clf.predict(X_test)
#pred = ['0']* len(y_test)
score = metrics.accuracy_score(y_test, pred)
print("accuracy: %0.3f" % score)
total.append(score)
results = {
'problem': [],
'method': [],
'score': []
}
if len(sys.argv) > 1 and sys.argv[1] == '--skip-train':
results = pd.read_csv("./data/results.csv")
else:
for classification_dataset in classification_dataset_names:
print("Starting", classification_dataset)
X, y = fetch_data(classification_dataset, return_X_y=True, local_cache_dir='./data/')
train_X, test_X, train_y, test_y = train_test_split(X, y)
rf = RandomForestClassifier()
lexRF = LexicaseForestClassifier()
rf.fit(train_X, train_y)
lexRF.fit(train_X, train_y)
rf_score = rf.score(test_X, test_y)
lexRF_score = lexRF.score(test_X, test_y)
results['problem'] = results['problem'] + ([classification_dataset] * 2)
results['method'] = results['method'] + ['RF', 'LexRF']
results['score'].append(rf_score)
results['score'].append(lexRF_score)
results = pd.DataFrame(results)
results.to_csv("./data/results.csv", index=False)