python matlplotlib/seaborn 绘制曲线的平均值标准差阴影图-Toy模板网

这篇具有很好参考价值的文章主要介绍了python matlplotlib/seaborn 绘制曲线的平均值标准差阴影图。希望对大家有所帮助。如果存在错误或未考虑完全的地方，请大家不吝赐教，您也可以点击"举报违法"按钮提交疑问。

1. seaborn

旧版本(0.8.1)中使用tsplot，新版本中使用lineplot
直线代表均值，阴影代表mean±std（带有置信区间，参数ci）

import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
sns.set()


def smooth(data, wd=2):
    """
    :param data: ndarray，一维或二维
    :param wd:
    :return:
    """
    if not (isinstance(wd, int) and wd > 0):
        raise ValueError('wd must be a positive integer')
    elif 1 == wd:
        return data
    else:
        weight = np.ones(wd) / wd
        if 1 == data.ndim:
            return np.convolve(weight, data, "same")
        elif 2 == data.ndim:
            smooth_data = []
            for d in data:
                d = np.convolve(weight, d, "same")
                smooth_data.append(d)
            return np.array(smooth_data)
        else:
            raise ValueError('data must be a one-dimensional or two-dimensional ndarray')


def get_data():
    returns1 = np.random.random((4, 100))        # 算法1，四个随机种子
    returns2 = np.random.random((4, 100)) + 1
    returns3 = np.random.random((4, 100)) + 2
    returns1 = smooth(returns1, 2)
    returns2 = smooth(returns2, 2)
    returns3 = smooth(returns3, 2)
    return returns1, returns2, returns3


np.random.seed(11)
data = get_data()
label = ['algo1', 'algo2', 'algo3']
df=[]
ax = range(10, 100+10)   # x轴刻度
for i in range(len(data)):
    df.append(pd.DataFrame(data[i], columns=ax).melt(var_name='episode',value_name='return'))
    df[i]['algo'] = label[i]
df=pd.concat(df, ignore_index=True)
# print(df)
sns.lineplot(x="episode", y="return", hue="algo", data=df)
plt.legend(loc='upper right')
# 'best', 'upper right', 'upper left', 'lower left', 'lower right',
#             'right', 'center left', 'center , right', 'lower center', 'upper center', 'center')
plt.title("")
plt.show()

#
# import seaborn as sns
# import matplotlib.pyplot as plt
# fmri = sns.load_dataset("fmri")
# fmri.head()
#
# sns.lineplot(data=fmri, x="timepoint", y="signal", hue="event")
# plt.show()