如何从MLPClassifier中适当地绘制(loss_curve_)获得的损失值

Question

在给定我的数据集的情况下,我使用以下代码通过MLPClassifier拟合模型:

tr_X, ts_X, tr_y, ts_y = train_test_split(X, y, train_size=.8)
model = MLPClassifier(hidden_layer_sizes=(32, 32),
              activation='relu',
              solver=adam,
              learning_rate='adaptive',
              early_stopping=True)

model.fit(tr_X, tr_y)
prd_r = model.predict(ts_X)
test_acc = accuracy_score(ts_y, prd_r) * 100.
loss_values = model.estimator.loss_curve_
print (loss_values)

如上所述,可以通过调用loss_curve_返回损失列表来获取每批的损失值.我懂了:

[0.69411586222116872, 0.6923803442491846, 0.66657293575365906, 0.43212054205535255, 0.23119813830216157, 0.15497928755966919, 0.11799652235604828, 0.095235784011297939, 0.079951427356068624, 0.069012741113626194, 0.061282868601098078, 0.054871864138797251, 0.049835046972801049, 0.046056362860260207, 0.042823979794540182, 0.040681220899240651, 0.038262366774481374, 0.036256840660697079, 0.034418333946277503, 0.033547227978657508, 0.03285581956914093, 0.031671266419493666, 0.030941451221456757]

我想绘制这些结果来代表loss curve这个模型.问题是,我不知道是什么x-axis,并y-axis会在这种情况下.如果我做y-axis这些损失值,x-axis这里应该显示损失曲线是减少还是增加？

任何提示或想法都表示赞赏.

Answer 1

该plot()命令过载,不需要x轴.如果您只是传入loss_curve_,则默认的x轴将是绘制的y值列表中的相应索引.例如,如果我们运行

import matplotlib.pyplot as plt

plt.plot(loss_values)
plt.show()

然后我们得到以下图表:

Answer 2

让我们根据它的迭代时间给出一个关于绘制损失和准确性的演示。

import numpy as np
import matplotlib.pyplot as plt
def draw_result(lst_iter, lst_loss, lst_acc, title):
    plt.plot(lst_iter, lst_loss, '-b', label='loss')
    plt.plot(lst_iter, lst_acc, '-r', label='accuracy')

    plt.xlabel("n iteration")
    plt.legend(loc='upper left')
    plt.title(title)

    # save image
    plt.savefig(title+".png")  # should before show method

    # show
    plt.show()


def test_draw():
    # iteration num
    lst_iter = range(100)

    # loss of iteration
    lst_loss = [0.01 * i + 0.01 * i ** 2 for i in xrange(100)]
    # lst_loss = np.random.randn(1, 100).reshape((100, ))

    # accuracy of iteration
    lst_acc = [0.01 * i - 0.01 * i ** 2 for i in xrange(100)]
    # lst_acc = np.random.randn(1, 100).reshape((100, ))
    draw_result(lst_iter, lst_loss, lst_acc, "sgd_method")


if __name__ == '__main__':
    test_draw()

输出如下：