位置：首页-资讯-后端开发

【python】plt.plot() 函数绘制loss曲线

2023-09-06 09:49

短信预约 -IT技能 免费直播动态提醒

一. 数据形式（输入数据）

二. 绘图（完整代码）

三. plt.plot() 函数（调整图形）

1. plt.plot(x, y)

2. plt.plot(x, y, "格式控制字符串")

2.1 "颜色"与"线型"

2.2 "点型"

3. plt.plot(x, y, "格式控制字符串", 关键字=参数)

一. 数据形式（输入数据）

训练过程中每个epoch都输出当前轮结果，输出数据保存在.txt文件，形式如下：

因为只是举个例子，只用30张图跑了5个epoch，不过数值不重要！过程先搞明白。

#每个epoch都输出当前轮结果print("epoch[%d/%d],train_loss,%.4f,train_acc,%.4f,train_miou,%.4f,eval_loss,%.4f,eval_acc,%.4f,eval_miou,%.4f,lr,%.6f,time,%ds" %              (epoch + 1, EPOCHES, epoch_loss, epoch_acc, epoch_miou, val_loss, val_acc, val_miou, learning_rate, time.time() - st_epoch))#输出结果形式epoch[1/5],train_loss,3.0900,train_acc,0.0190,train_miou,0.0046,eval_loss,3.0398,eval_acc,0.0438,eval_miou,0.0075,lr,0.000005,time,23sepoch[2/5],train_loss,2.9437,train_acc,0.0667,train_miou,0.0090,eval_loss,2.9367,eval_acc,0.1133,eval_miou,0.0082,lr,0.000005,time,25sepoch[3/5],train_loss,2.8345,train_acc,0.1875,train_miou,0.0141,eval_loss,2.8379,eval_acc,0.2588,eval_miou,0.0087,lr,0.000005,time,25sepoch[4/5],train_loss,2.7256,train_acc,0.3285,train_miou,0.0163,eval_loss,2.7290,eval_acc,0.4454,eval_miou,0.0072,lr,0.000005,time,24sepoch[5/5],train_loss,2.6142,train_acc,0.5298,train_miou,0.0167,eval_loss,2.6255,eval_acc,0.5811,eval_miou,0.0032,lr,0.000005,time,27s

二. 绘图（完整代码）

逐行读取数据
用split()按所给符号分割
把所需列的数据按索引添加到对应的表里
用plt画图，需要什么形式的图，函数的参数可以参考第三部分函数介绍自行更改

#完整代码import matplotlib.pyplot as pltfile = open('log.txt')  #打开文档lines = file.readlines() #读取文档数据#epoch = list(1, range(len(lines))+1) #epoch可以直接赋值，不放心的就用下面epoch的代码epoch = []train_loss = []val_loss = []for line in lines:# split用于将每一行数据用自定义的符号（我用的是逗号）分割成多个对象    # 取分割后的第0列，转换成float格式后添加到epoch列表中    epoch.append(str(line.split(',')[0]))    # 取分割后的第2列，转换成float格式后添加到train_loss列表中    train_loss.append(float(line.split(',')[2]))    #取分割后的第8列，转换成float格式后添加到val_loss列表中    val_loss.append(float(line.split(',')[8]))plt.figure()         plt.title('loss during training')  #标题plt.plot(epoch, train_loss, label="train_loss")plt.plot(epoch, val_loss, label="valid_loss")plt.legend()plt.grid()plt.show()

输出结果：

三. plt.plot() 函数（调整图形）

matplotlib.pyplot模块下的一个函数, 用于画图。它可以绘制点和线, 并且对其样式进行控制。

函数定义为plt.plot(*args, **kwargs)

import matplotlib.pyplot as plthelp(plt.plot) # 查看英文函数定义

*args, 可变位置参数, 以元组形式存放了很多无名参数.
**kwargs, 可变关键字参数, 以字典形式存放了很多关键字及参数. 调用时可传入
*args要放在**kwargs之前.
本函数的*args, 允许传入多对x和y和一个可选的"格式控制字符串".
本函数的**kwargs, 允许传入多个可选的关键字参数.

1. plt.plot(x, y)

x为x轴数据, y为y轴数据。
x, y可传入(元组), [列表], np.array, pd.Series。
也可同时传入多组。

import matplotlib.pyplot as pltimport numpy as npimport pandas as pd#示例一：x为x轴数据, y为y轴数据x=[3,4,5] # [列表]y=[2,3,2] # x,y元素个数N应相同plt.plot(x,y)plt.show()#示例二：x, y可传入(元组), [列表], np.array, pd.Seriesx=(3,4,5) # (元组)y1=np.array([3,4,3]) # np.arrayy2=pd.Series([4,5,4]) # pd.Seriesplt.plot(x,y1)plt.plot(y2)  # x可省略,默认[0,1..,N-1]递增plt.show() # plt.show()前可加多个plt.plot(),画在同一张图上#示例三：可传入多组x, yx=(3,4,5)y1=np.array([3,4,3])y2=pd.Series([4,5,4])plt.plot(x,y1,x,y2) # 此时x不可省略plt.show()

结果示例：

2. plt.plot(x, y, "格式控制字符串")

点和线的格式可以用"格式控制字符串"设置，"最多可以包括三部分, "颜色", "点型", "线型"。

2.1 "颜色"与"线型"

如果只控制"颜色", 格式控制字符串还可以输入英文全称, 如"red", 甚至是十六进制RGB字符串, 如"#FF0000". python可用的"颜色"大全

    =============    ===============================    character        color    =============    ===============================    ``'b'``          blue 蓝    ``'g'``          green 绿    ``'r'``          red 红    ``'c'``          cyan 蓝绿    ``'m'``          magenta 洋红    ``'y'``          yellow 黄    ``'k'``          black 黑    ``'w'``          white 白    =============    ===============================

    =============    ===============================    character        description    =============    ===============================    ``'-'``          solid line style 实线    ``'--'``         dashed line style 虚线    ``'-.'``         dash-dot line style 点画线    ``':'``          dotted line style 点线    =============    ===============================

#示例import numpy as npimport pandas as pdimport matplotlib.pyplot as pltcolor=['b','g','r','c','m','y','k','w']linestyle=['-','--','-.',':']dic1=[[0,1,2],[3,4,5]]x=pd.DataFrame(dic1)dic2=[[2,3,2],[3,4,3],[4,5,4],[5,6,5]]y=pd.DataFrame(dic2)# 循环输出所有"颜色"与"线型"for i in range(2):    for j in range(4):        plt.plot(x.loc[i],y.loc[j],color[i*4+j]+linestyle[j]) plt.show()

输出结果：

2.2 "点型"

    =============    ===============================    character        description    =============    ===============================    ``'.'``          point marker    ``','``          pixel marker    ``'o'``          circle marker    ``'v'``          triangle_down marker    ``'^'``          triangle_up marker    ``'<'``          triangle_left marker    ``'>'``          triangle_right marker    ``'1'``          tri_down marker    ``'2'``          tri_up marker    ``'3'``          tri_left marker    ``'4'``          tri_right marker    ``'s'``          square marker    ``'p'``          pentagon marker    ``'*'``          star marker    ``'h'``          hexagon1 marker    ``'H'``          hexagon2 marker    ``'+'``          plus marker    ``'x'``          x marker    ``'D'``          diamond marker    ``'d'``          thin_diamond marker    ``'|'``          vline marker    ``'_'``          hline marker    =============    ===============================

#示例import numpy as npimport pandas as pdimport matplotlib.pyplot as pltmarker=['.',',','o','v','^','<','>','1','2','3','4','s','p','*','h','H','+','x','D','d','|','_','.',',']dic1=[[0,1,2],[3,4,5],[6,7,8],[9,10,11],[12,13,14],[15,16,17]]x=pd.DataFrame(dic1)dic2=[[2,3,2.5],[3,4,3.5],[4,5,4.5],[5,6,5.5]]y=pd.DataFrame(dic2)# 循环输出所有"点型"for i in range(6):    for j in range(4):        plt.plot(x.loc[i],y.loc[j],"b"+marker[i*4+j]+":") # "b"蓝色,":"点线plt.show()

3. plt.plot(x, y, "格式控制字符串", 关键字=参数)

本函数的**kwargs, 允许传入多个可选的关键字参数

#示例y=[2,3,2,4,5] # 青色,线宽10,星星,点尺寸50,点填充绿色,点边缘宽度6,点边缘青色plt.plot(y,color="c",linewidth=10,marker="*",markersize=20,         markerfacecolor="g",markeredgewidth=3,markeredgecolor="c")plt.show()