def check_code(mode, gru_mode):
if(mode == 'simple'):
train_df = pd.read_csv('/home/Team4/Team4/dataset/train_percentile.csv')
test_df = pd.read_csv('/home/Team4/Team4/dataset/testB_percentile.csv')
train_add = pd.read_csv('/home/Team4/Team4/dataset/train_old_wind_4240.csv')
testA_add = pd.read_csv('/home/Team4/Team4/dataset/testB_old_wind_4240.csv')
train_1ave8extend = pd.read_csv('/home/Team4/Team4/dataset/train_new_wind_1ave_8extend.csv')
test_1ave = pd.read_csv('/home/Team4/Team4/dataset/testB_new_wind_1ave_8extend.csv')
else:
trainfile = '/home/Team4/CIKM2017/train.txt'
testBfile = '/home/Team4/CIKM2017/testB.txt'
#生成训练集数据,老的风
train_add = dp.dataprocess(trainfile, data_type='train', windversion='old')
#生成测试集B数据,老的风
testA_add = dp.dataprocess(testBfile, data_type='testB', windversion='old')
#生成训练集数据,1ave8extend
train_1ave8extend = dp.dataprocess(trainfile, data_type='train', windversion='new')
#生成测试集B数据,1ave
test_1ave = dp.dataprocess(testBfile, data_type='testB', windversion='new')
#生成训练集数据
train_df = gp.data_process(trainfile, data_type='train')
#生成测试集B数据
test_df = gp.data_process(testBfile, data_type='testB')
print('#data process has been done')
result_xgb = xgbm.xgbmodeltrain(train_1ave8extend, test_1ave)
print('#xgb model has been done')
index = fs.pre_train(train_df=train_df, test_df=test_df, train_add=train_add, test_add=testA_add)
valid = rf.rf_model(train_df, test_df, 'train', train_add, testA_add, ne=100)
ne = 1100
result_rf = rf.rf_model(train_df, test_df, 'trai', train_add, testA_add, ne, index=index)
print('#rf model has been done')
result_bigru = bigru.BiGRU_train(train_df, test_df, valid, gru_mode).reshape(2000)
print('#bigru model has been done')
ensemble = (result_xgb+result_rf+result_bigru)/3.0
np.savetxt("/home/Team4/Team4/result/submit_Team4.csv", ensemble)
def check_code(mode, gru_mode):
if mode == 'simple':
train_df = pd.read_csv('/data/yuyang/weather/data/data_shenzhen/data_processed/train_percentile.csv')
test_df = pd.read_csv('/data/yuyang/weather/data/data_shenzhen/data_processed/testB_percentile.csv')
train_add = pd.read_csv('/data/yuyang/weather/data/data_shenzhen/data_processed/train_old_wind_4240.csv')
testA_add = pd.read_csv('/data/yuyang/weather/data/data_shenzhen/data_processed/testB_old_wind_4240.csv')
train_1ave8extend = pd.read_csv('/data/yuyang/weather/data/data_shenzhen/data_processed/train_new_wind_1ave_8extend.csv')
test_1ave = pd.read_csv('/data/yuyang/weather/data/data_shenzhen/data_processed/testB_new_wind_1ave_8extend.csv')
else:
trainfile = '/data/yuyang/weather/data/data_shenzhen/CIKM2017_train/train'
testBfile = '/data/yuyang/weather/data/data_shenzhen/CIKM2017_testB/CIKM2017_testB/test'
args = ['00', '01', '02', '03', '04', '05', '06', '07', '08', '09', '10', '11', '12', '13', '14']
p = Pool(15)
# 生成训练集数据,老的风
# train_add = dp.dataprocess(trainfile, data_type='train', windversion='old')
print('start----dp-train-old')
func_dp_train_old = partial(dp.dataprocess, trainfile, 'train', 'old')
p.map(func_dp_train_old, args)
# 生成测试集B数据,老的风
# testA_add = dp.dataprocess(testBfile, data_type='testB', windversion='old')
print('start----dp-test-old')
func_dp_test_old = partial(dp.dataprocess, testBfile, 'testB', 'old')
p.map(func_dp_test_old, args)
# 生成训练集数据,1ave8extend
# train_1ave8extend = dp.dataprocess(trainfile, data_type='train', windversion='new')
print('start----dp-train-new')
func_dp_train_new = partial(dp.dataprocess, trainfile, 'train', 'new')
p.map(func_dp_train_new, args)
# 生成测试集B数据,1ave
# test_1ave = dp.dataprocess(testBfile, data_type='testB', windversion='new')
print('start----dp-testB-new')
func_dp_test_new = partial(dp.dataprocess, testBfile, 'testB', 'new')
p.map(func_dp_test_new, args)
# 生成训练集数据
# train_df = gp.data_process(trainfile, data_type='train')
print('start----gp-train')
func_gp_train = partial(gp.data_process, trainfile, 'train')
p.map(func_gp_train, args)
# 生成测试集B数据
# test_df = gp.data_process(testBfile, data_type='testB')
print('start----gp-test')
func_gp_test = partial(gp.data_process, testBfile, 'testB')
p.map(func_gp_test, args)
print('data process has been done')
return 0
print('#data process has been done')
daytime = time.strftime('%Y-%m-%d-%H', time.localtime(time.time()))
print(daytime)
result_xgb = xgbm.xgbmodeltrain(train_1ave8extend, test_1ave)
np.savetxt("/data/yuyang/weather/result/shenzhen/" + "xgb_pre-{}.csv".format(daytime), result_xgb)
print('#xgb model has been done')
index = fs.pre_train(train_df=train_df, test_df=test_df, train_add=train_add, test_add=testA_add)
valid = rf.rf_model(train_df, test_df, 'train', train_add, testA_add, ne=100)
ne = 1100
result_rf = rf.rf_model(train_df, test_df, 'trai', train_add, testA_add, ne, index=index)
np.savetxt("/data/yuyang/weather/result/shenzhen/" + "rf_pre-{}.csv".format(daytime), result_rf)
print('#rf model has been done')
result_bigru = bigru.BiGRU_train(train_df, test_df, valid, gru_mode).reshape(2000)
np.savetxt("/data/yuyang/weather/result/shenzhen/" + "bi-gru_pre-{}.csv".format(daytime), result_bigru)
print('#bigru model has been done')
ensemble = (result_xgb + result_rf + result_bigru) / 3.0
np.savetxt("/data/yuyang/weather/result/shenzhen/" + "ensemble-{}.csv".format(daytime), ensemble)