时间戳

首页 > 解决方案 > 当我使用 XGBOOST 回归时，max_depth 的不同值只能得到负分

python - 当我使用 XGBOOST 回归时，max_depth 的不同值只能得到负分

问题描述

我的代码运行良好，但问题是所有的 train_scores 和 test_scores 都是负值。如果您能帮助我找出代码中的问题，我会很高兴。我希望评估模型在不同 max_depth 值下的性能：1、3、5、6、7、9、11、13、15、17、19。

这是我得到的输出：

train_scores = [-1.53​​30488137183185, -1.253742668302357, -1.212924006705376, -1.209022226685474, -1.2087753262347243, -1.2087753262347243, -1.2087753262347243, -1.2087753262347243, -1.2087753262347243, -1.2087753262347243]

test_scores = [-1.7155087611718076, -1.4551497092616446, -1.427034811918214, -1.4260848404210638, -1.42582407694383, -1.42582407694383, -1.42582407694383, -1.42582407694383, -1.42582407694383, -1.42582407694383]

这是我的代码：

mse_train1, mse_test1,max_depth1, RMSE_Train1, RMSE_Test1, train_scores, test_scores = list(), list(), list(), list(),list(), list(),list()

model = XGBRegressor(learning_rate= 0.04,max_depth= 1,n_estimators= 40,subsample= 0.7)

# Training and Evaluate the model for the dataframe contains important features
# -----------------------------------------------------------------------------

for iter in range(1, 20, 2):
    
    max_depth1.append(iter)
    
    model.fit(train_set, train_set_RET)
    y_train_predicted = model.predict(train_set)
    train_score =  model.score(train_set, train_set_RET)
    train_scores.append(train_score)
    
    y_test_predicted = model.predict(test_set)
    test_score =  model.score(test_set, test_set_RET)
    test_scores.append(test_score)    

    mse_train = mean_squared_error(train_set_RET, y_train_predicted)
    mse_train1.append(mse_train)
    RMSE_Train = np.sqrt(mse_train)
    RMSE_Train1.append (RMSE_Train)

    mse_test = mean_squared_error(test_set_RET, y_test_predicted)
    mse_test1.append(mse_test)
    RMSE_Test = np.sqrt(mse_test)
    RMSE_Test1.append (RMSE_Test)
    #print('>%d, train: %.3f, test: %.3f' % (i, train_score, test_score))
    #print("Iteration: {} Train mse: {} Test mse: {}".format(iter, mse_train, mse_test))
    model.max_depth += 2
    
def plot1_fun (num_trees,mse_train1, mse_test1):
    
    pyplot.plot(num_trees, mse_train1, marker='.', label= 'MSE on Train Data')
    pyplot.plot(num_trees, mse_test1, marker='.', label= 'MSE on Test Data')

    # axis labels
    pyplot.xlabel('max depth')
    pyplot.ylabel('Mean Squared Error')
    
    # show the legend
    pyplot.legend()
    
    # show the plot
    pyplot.show()


plot1_fun(num_trees,mse_train1, mse_test1)

print ('train_scores =', train_scores)
print ('test_scores =', test_scores)

标签： pythonxgboost

解决方案

推荐阅读