In [47]:

    

import pandas as pd
import numpy as np
import xgboost as xgb


    

In [48]:

    

path = "/home/siyuan/Documents/Boya/bimbo_kaggle_reformat-master"


    

In [49]:

    

# hold_data is for for parameter tuining, already got them
train_data = pd.read_csv(path + "/data/processed/train.csv")
train_label = pd.read_csv(path + "/data/processed/train_label.csv")
test_data = pd.read_csv(path + "/data/processed/test_label.csv")


    

In [50]:

    

test_data.columns = train_data.columns


    

In [51]:

    

print train_data.shape, train_label.shape


    

    




(13049, 401) (13049, 1)

In [52]:

    

param = {'booster':'gbtree',
         'nthread': 10,
         'max_depth':5, 
         'eta':0.2,
         'silent':1,
         'subsample':0.7, 
         'objective':'reg:linear',
         'eval_metric':'rmse',
         'colsample_bytree':0.7}


    

In [53]:

    

num_round = 566
dtrain = xgb.DMatrix(train_data, label = train_label, missing= np.nan)
bst_bimbo = xgb.train(param, dtrain, num_round)
print 'training finished!'

bst_bimbo.save_model(path + '/models/xgb_bimbo.model')

dtest = xgb.DMatrix(test_data, missing= np.nan)
submission_bimbo = bst_bimbo.predict(dtest)
submission_bimbo.to_csv("sample_sub.csv", index = False)
print 'predicting finished!'


    

    




training finished!
predicting finished!

In [ ]: