lightGBM

from lightgbm import LGBMRegressor
from xgboost import XGBRegressor
import numpy as np
import pandas as pd

train = pd.read_csv('./zhengqi_train.txt',sep = '\t')
test = pd.read_csv('./zhengqi_test.txt',sep = '\t')

X_train = train.iloc[:,:-1]
y_train = train['target']
X_train.shape

%%time
light = LGBMRegressor()
light.fit(X_train,y_train)
y_ = light.predict(test)
pd.Series(y_).to_csv('./light.txt',index = False)

%%time
xgb = XGBRegressor(max_depth = 50,n_estimators = 100)
xgb.fit(X_train,y_train)
y_ = xgb.predict(test)
pd.Series(y_).to_csv('./xgb.txt',index = False)

#协方差，两个属性之间的关系，协方差绝对值越大，两个属性之间的关系越密切
cov = train.cov()
#目标值和属性之间的关系
cov

#协方差较小，定义为不太重要的参数
drop_labels = cov.index[cov.loc['target'].abs()<0.1]#abs绝对值
drop_labels.shape

#把不重要属性去除

X_train.drop(drop_labels,axis = 1,inplace = True)
X_train.shape

light = LGBMRegressor()
light.fit(X_train,y_train)
y_ = light.predict(test)
pd.Series(y_).to_csv('./light2.txt',index = False)

X_test = pd.read_csv('./zhengqi_test.txt',sep = '\t')

#删除样本不均匀的特征
drop_labels = ["V5","V9","V11"]

X_train = train.iloc[:,0:-1]
X_test = pd.read_csv('./zhengqi_test.txt',sep = '\t')
X_train.drop(drop_labels,axis = 1,inplace = True)
X_test.drop(drop_labels,axis = 1,inplace = True)

X_test.shape

light = LGBMRegressor()
light.fit(X_train,y_train)
y_ = light.predict(X_test)
pd.Series(y_).to_csv('./light3.txt',index = False)