Regression Example with XGBRegressor in Python

   XGBoost stands for "Extreme Gradient Boosting" and it is an implementation of gradient boosting machines. The XGBoost is a popular supervised machine learning model with characteristics like fast in computation, parallelization, and better performance. You can find more about the model in this link. In this post, we'll learn how to define the XGBRegressor model and predict regression data in Python.
   The tutorial covers:
  1. Preparing data
  2. Defining and fitting the model
  3. Predicting and checking the results
   We'll start by loading the required libraries. You may need to install them if they are not available on your machine.

import xgboost as xgb
from sklearn.datasets import load_boston
from sklearn.model_selection import train_test_split
from sklearn.model_selection import cross_val_score, KFold
from sklearn.metrics import mean_squared_error
import matplotlib.pyplot as plt
import numpy as np 


Preparing data

   We use Boston house-price dataset as regression dataset in this tutorial. After loading the dataset, first, we'll separate data into x - feature and y - label. Then we'll split them into the train and test parts. Here, I'll extract 15 percent of the dataset as test data.

boston = load_boston()
x, y = boston.data, boston.target
xtrain, xtest, ytrain, ytest=train_test_split(x, y, test_size=0.15)


Defining and fitting the model

   For the regression problem, we'll use XGBRegressor class of the xgboost package and we can define it with its default parameters. You can also set the new parameter values according to your data characteristics.

xgbr = xgb.XGBRegressor() 
print(xgbr)
XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,
       colsample_bynode=1, colsample_bytree=1, gamma=0,
       importance_type='gain', learning_rate=0.1, max_delta_step=0,
       max_depth=3, min_child_weight=1, missing=None, n_estimators=100,
       n_jobs=1, nthread=None, objective='reg:linear', random_state=0,
       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,
       silent=None, subsample=1, verbosity=1)

Next, we'll fit the model with train data.

xgbr.fit(xtrain, ytrain)


Predicting and checking the results



After training the model, we'll check the model accuracy with cross-validation method.

scores = cross_val_score(xgbr, xtrain,ytrain,cv=5)
print("Mean cross-validation score: %.2f" % scores.mean())
Mean cross-validataion score: 0.87 

Cross-validation with a k-fold method can be checked as a following.

kfold = KFold(n_splits=10, shuffle=True)
kf_cv_scores = cross_val_score(xgbr, xtrain, ytrain, cv=kfold )
print("K-fold CV average score: %.2f" % kf_cv_scores.mean())
K-fold CV average score: 0.87

Both methods show that the model is around 88 %  accurate on average.
Next, we can predict test data and check its accuracy. Here, we'll use MSE and RMSE as accuracy metrics.

ypred = xgbr.predict(xtest)
mse = mean_squared_error(ytest,ypred)
print("MSE: %.2f" % mse)
MSE: 3.35
print("RMSE: %.2f" % np.sqrt(mse))
RMSE: 1.83 


Finally, we'll visualize the original and predicted test data in a plot.

x_ax = range(len(ytest))
plt.scatter(x_ax, ytest, s=5, color="blue", label="original")
plt.plot(x_ax, ypred, lw=0.8, color="red", label="predicted")
plt.legend()
plt.show()




   In this post, we've briefly learned how to use XGBRegressor to predict regression data in Python. Thank you for reading.
   The full source code is listed below.

import xgboost as xgb
from sklearn.datasets import load_boston
from sklearn.model_selection import train_test_split
from sklearn.model_selection import cross_val_score, KFold
from sklearn.metrics import mean_squared_error
import matplotlib.pyplot as plt 
import numpy as np
 
boston = load_boston()
x, y = boston.data, boston.target
xtrain, xtest, ytrain, ytest=train_test_split(x, y, test_size=0.15)

xgbr = xgb.XGBRegressor()
print(xgbr)

xgbr.fit(xtrain, ytrain)
 
# - cross validataion 
scores = cross_val_score(xgbr, xtrain, ytrain, cv=5)
print("Mean cross-validation score: %.2f" % scores.mean())

kfold = KFold(n_splits=10, shuffle=True)
kf_cv_scores = cross_val_score(xgbr, xtrain, ytrain, cv=kfold )
print("K-fold CV average score: %.2f" % kf_cv_scores.mean())
 
ypred = xgbr.predict(xtest)
mse = mean_squared_error(ytest, ypred)
print("MSE: %.2f" % mse)
print("RMSE: %.2f" % np.sqrt(mse))

x_ax = range(len(ytest))
plt.scatter(x_ax, ytest, s=5, color="blue", label="original")
plt.plot(x_ax, ypred, lw=0.8, color="red", label="predicted")
plt.legend()
plt.show()



3 comments:

  1. Hello,
    I've a couple of question.
    1. What are labels for x and y axis in the above graph?

    2. Then I’m trying to understand the following example.
    I’m confused about the first piece of code. It seems to me that cross-validation and Cross-validation with a k-fold method are performing the same actions. In the second example just 10 times more. The result is the same. I dont understand the cross-validation in first example what is for?
    Thanks,
    Marco

    ReplyDelete
  2. Hi,
    1. The plot describes 'medv' column of boston dataset (original and predicted). x label is the number of sample and y label is the value of 'medv'
    2. They explain two ways of implementaion of cross-validation. You can use one of them.


    ReplyDelete
  3. how can write python code to upload similar work done like this in order to submit on kaggle.com. Thanks

    ReplyDelete