def get_reslut(d_to_str, time, storeId, pred):
args = dict()
args['storeId'] = storeId
args['starttime'] = time
args['preds'] = pred # 人流客流特征参数 'traffic'/'customer'
args['day'] = d_to_str
args_pref_features = data_helper.data_loader(train=False, args=args)
model = joblib.load('train_model.m')
return model.predict([args_pref_features])
def run():
args = dict()
args['storeId'] = 5
args['starttime'] = 11
args['preds'] = 'customer' #人流客流特征参数 'traffic'/'customer'
args['day'] = '2017-08-08'
args_pref_features = data_helper.data_loader(train=False, args=args)
model = joblib.load('train_model.m')
print model.predict([args_pref_features])
def get_reslut(d_to_str, time, storeId, pred,model):
args = dict()
args['storeId'] = storeId
args['starttime'] = time
args['preds'] = pred # 人流客流特征参数 'traffic'/'customer'
args['day'] = d_to_str
try :
args_pref_features = data_helper.data_loader(train=False, args=args)
res=model.predict([args_pref_features])
except BaseException as e:
app.logger.error(e)
return res
def train_task():
#load train data
train_X_data, train_y_data = data_helper.data_loader()
#split the train data and test data
X_train, X_test, y_train, y_test = train_test_split(train_X_data, train_y_data, \
test_size=0.25, random_state=33)
# new model and train model
GBR = GradientBoostingRegressor()
GBR = GBR.fit(X_train, y_train)
#save model
joblib.dump(GBR, "train_model.m")
def train_task():
#load train data
try:
train_X_data, train_y_data, storeId = data_helper.data_loader()
except BaseException as e:
logger.error(e)
#保存下来所有店铺id
pkl.dump(storeId, open('storeId.pkl', 'wb'))
#train model
try:
GBR = GradientBoostingRegressor()
GBR = GBR.fit(train_X_data, train_y_data)
except BaseException as e:
logger.error(e)
#save model
joblib.dump(GBR, "train_model.m")
'The results dir including log, model, vocabulary and some images.(default=./results/)'
)
parser.add_argument('--data_path',
default='data/baidu_95.csv',
type=str,
help='data path')
parser.add_argument('--vocab_save_dir',
default='data/',
type=str,
help='data path')
parser.add_argument('--workers',
default=32,
type=int,
help='use worker count')
params = parser.parse_args()
print('Parameters:', params, '\n')
if not os.path.exists(params.results_dir):
os.mkdir(params.results_dir)
timestamp = time.strftime("%Y-%m-%d-%H-%M", time.localtime(time.time()))
os.mkdir(os.path.join(params.results_dir, timestamp))
os.mkdir(os.path.join(params.results_dir, timestamp, 'log/'))
X_train, X_test, y_train, y_test = data_loader(params)
train(X_train, X_test, y_train, y_test, params,
os.path.join(params.results_dir, 'TextCNN.h5'))
# -*- coding: utf-8 -*-
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.cross_validation import train_test_split
from sklearn.externals import joblib
import data_helper
# update mysql.user set authentication_string=password("123456") where user="******";
# UPDATE mysql.user SET Password=PASSWORD("123456") WHERE User='******' and Host='localhost';
if __name__ == '__main__':
#load train data
# df=data_helper.data_loader()
# print df
train_X_data,train_y_data=data_helper.data_loader()
#split the train data and test data
X_train, X_test, y_train, y_test = train_test_split(train_X_data, train_y_data, \
test_size=0.25, random_state=33)
# new model and train model
GBR=GradientBoostingRegressor()
GBR=GBR.fit(X_train, y_train)
#save model
joblib.dump(GBR, "train_model.m")
pred=GBR.predict(X_test)
#test the model score
result=zip(y_test,pred)
print "real_value\tpred_value"
import numpy as np
from sklearn.linear_model import SGDRegressor
from sklearn.neural_network import MLPRegressor
from sklearn.ensemble import ExtraTreesClassifier
from sklearn.svm import SVR
from data_helper import data_loader, mae, model_saver
train_x, train_y, test_x, test_y = data_loader()
train_y = np.array(train_y)
test_y = np.array(test_y)
train_y_in, train_y_out = train_y[:, 0], train_y[:, 1]
test_y_in, test_y_out = test_y[:, 0], test_y[:, 1]
def svr_model():
svr_rbf = SVR(kernel='rbf', C=100, gamma=0.1, epsilon=.1, verbose=True)
svr_rbf.fit(train_x, train_y_in)
pre_y = svr_rbf.predict(test_x)
print(mae(np.array(pre_y), test_y_in))
def sgd_model():
sgd = SGDRegressor(loss='huber',
max_iter=1000000,
learning_rate='invscaling',
eta0=20,
verbose=True)
sgd.fit(train_x, train_y_in)
pre_y = sgd.predict(test_x)
print(mae(np.array(pre_y), test_y_in))