In [21]:

    

import datetime
import numpy as np
import pandas as pd
import xgboost as xgb
from sklearn import preprocessing
from sklearn.metrics import r2_score
import matplotlib.pyplot as plt
plt.style.use("ggplot")
%matplotlib inline


    

In [6]:

    

train = pd.read_csv("data/train.csv", index_col="ID")
test = pd.read_csv("data/test.csv", index_col="ID")


    

In [7]:

    

y_train = train["y"]
x_train = train.drop(["y"], axis=1)

for c in x_train.columns:
    if x_train[c].dtype == "object":
        lbl = preprocessing.LabelEncoder()
        lbl.fit(list(x_train[c].values)) 
        x_train[c] = lbl.transform(list(x_train[c].values))

y_train = y_train.astype(np.float32)
x_train = x_train.astype(np.float32)


    

In [11]:

    

x_test = test.copy()

for c in x_test.columns:
    if x_test[c].dtype == "object":
        lbl = preprocessing.LabelEncoder()
        lbl.fit(list(x_test[c].values)) 
        x_test[c] = lbl.transform(list(x_test[c].values))
        
x_test = x_test.astype(np.float32)


    

In [13]:

    

xgb_params = {
    "eta": 0.05,
    "max_depth": 5,
    "subsample": 0.7,
    "colsample_bytree": 0.7,
    "objective": "reg:linear",
    "eval_metric": "rmse",
    "silent": 1,
    "seed":42
}


    

In [14]:

    

dtrain = xgb.DMatrix(x_train, y_train)
dtest = xgb.DMatrix(x_test)


    

In [15]:

    

cv_output = xgb.cv(xgb_params, dtrain, num_boost_round=1000, early_stopping_rounds=20, 
                   verbose_eval=50, show_stdv=False)


    

    




[0]	train-rmse:95.968	test-rmse:95.9667
[50]	train-rmse:10.9934	test-rmse:11.1884
[100]	train-rmse:7.48883	test-rmse:8.39515

In [16]:

    

cv_output[["train-rmse-mean", "test-rmse-mean"]].plot()
pass


    

In [17]:

    

num_boost_rounds = len(cv_output)
model = xgb.train(dict(xgb_params, silent=0), dtrain, num_boost_round=num_boost_rounds)


    

In [18]:

    

xgb.plot_importance(model, max_num_features=20)


    

    
Out[18]:





<matplotlib.axes._subplots.AxesSubplot at 0x11058f8d0>






    

In [22]:

    

y_pred = model.predict(dtrain)


    

In [23]:

    

r2_score(y_train, y_pred)


    

    
Out[23]:





0.63256989950379339

In [29]:

    

y_predict = model.predict(dtest)


    

In [30]:

    

output = pd.DataFrame({"ID": x_test.index, "y": y_predict})


    

In [32]:

    

output.head()


    

    
Out[32]:






  

    

      

      
ID
      
y
    
  
  

    

      
0
      
1
      
81.746040
    
    

      
1
      
2
      
98.966293
    
    

      
2
      
3
      
82.673309
    
    

      
3
      
4
      
77.018295
    
    

      
4
      
5
      
109.772110
    
  

In [33]:

    

output.to_csv("submissions_{}.csv".format(datetime.datetime.today()), index=False)


    

In [ ]:

    




    

In [ ]: