fantastic-matplotlib:案例集合：

画四个指标的learning curve

# -*- coding: utf-8 -*-
"""
Created on Sat Dec  5 09:42:05 2020

@author: 姜金
"""

import matplotlib.pyplot as plt
import numpy as np
import matplotlib.gridspec as gridspec
from matplotlib.ticker import FuncFormatter

def to_percent(temp, position):
    return '%1.0f'%(10*temp) + '%'

p_10 = [0.5490909090909091, 348, 294, 202, 256]
p_20 = [0.5854545454545454, 443, 349, 107, 201]
p_30 = [0.7045454545454546, 489, 264, 61, 286]
p_40 = [0.7145454545454546, 494, 258, 56, 292]
p_50 = [0.6727272727272727, 508, 318, 42, 232]
p_60 = [0.7363636363636363, 511, 251, 39, 299]
p_70 = [0.76, 515, 229, 35, 321]
p_80 = [0.7754545454545455, 497, 194, 53, 356]
p_90 = [0.7781818181818182, 499, 193, 51, 357]
p_100 = [0.769090909090909, 502, 206, 48, 344]
#5 EPOCH

def metrics_comput(p):
    acc = p[0]
    TP = p[1]
    FP = p[2]
    FN = p[3]
#    TN = p[4]
    P = TP/(TP+FP)
    R = TP/(TP+FN)
    F1 = 2*P*R/(P+R)
    return acc, P, R, F1



#plt.figure()
plt.figure(figsize=(10,8))


x_data = np.arange(10)
gs = gridspec.GridSpec(2,2) #将绘图区分成两行三列
ax1 = plt.subplot(gs[0,0])#指定ax1占用第一行(0)整行
ax2 = plt.subplot(gs[0,1])#指定ax1占用第一行(0)整行
ax3 = plt.subplot(gs[1,0])#指定ax2占用第二行(1)的第一格(第二个参数为0)
ax4 = plt.subplot(gs[1,1])#指定ax3占用第二行(1)的第二、三格(第二个参数为1：3)

y1 = 100*np.array([metrics_comput(p_10)[0],metrics_comput(p_20)[0],metrics_comput(p_30)[0],metrics_comput(p_40)[0],metrics_comput(p_50)[0],metrics_comput(p_60)[0],metrics_comput(p_70)[0],metrics_comput(p_80)[0],metrics_comput(p_90)[0],metrics_comput(p_100)[0]])
y11 = np.array([1,2,3,4,5,6,7,8,9,10])*10
ax1.plot(x_data,y1,label='single')
ax1.plot(x_data,y11,label='pairing')
ax1.legend()
ax1.set_xlabel('#Percentage of sentences')
ax1.set_ylabel('Accuracy(%)')
ax1.xaxis.set_major_formatter(FuncFormatter(to_percent))


y2 = 100* np.array([metrics_comput(p_10)[1],metrics_comput(p_20)[1],metrics_comput(p_30)[1],metrics_comput(p_40)[1],metrics_comput(p_50)[1],metrics_comput(p_60)[1],metrics_comput(p_70)[1],metrics_comput(p_80)[1],metrics_comput(p_90)[1],metrics_comput(p_100)[1]])
y22 = np.array([1,2,3,4,5,6,7,8,9,10])*10
ax2.plot(x_data,y2,label='single')
ax2.plot(x_data,y22,label='pairing')
ax2.legend()
ax2.set_xlabel('#Percentage of sentences')
ax2.set_ylabel('Precision(%)')

ax2.xaxis.set_major_formatter(FuncFormatter(to_percent))

y3 = 100*np.array([metrics_comput(p_10)[2],metrics_comput(p_20)[2],metrics_comput(p_30)[2],metrics_comput(p_40)[2],metrics_comput(p_50)[2],metrics_comput(p_60)[2],metrics_comput(p_70)[2],metrics_comput(p_80)[2],metrics_comput(p_90)[2],metrics_comput(p_100)[2]])
y33 = np.array([1,2,3,4,5,6,7,8,9,10])*10
ax3.plot(x_data,y3,label='single')
ax3.plot(x_data,y33,label='pairing')
ax3.legend()
ax3.set_xlabel('#Percentage of sentences')
ax3.set_ylabel('Recall(%)')
ax3.xaxis.set_major_formatter(FuncFormatter(to_percent))

y4 = 100* np.array([metrics_comput(p_10)[3],metrics_comput(p_20)[3],metrics_comput(p_30)[3],metrics_comput(p_40)[3],metrics_comput(p_50)[3],metrics_comput(p_60)[3],metrics_comput(p_70)[3],metrics_comput(p_80)[3],metrics_comput(p_90)[3],metrics_comput(p_100)[3]])
y44 = np.array([1,2,3,4,5,6,7,8,9,10])*10
ax4.plot(x_data,y4,label='single')
ax4.plot(x_data,y44,label='pairing')
ax4.legend()
ax4.set_xlabel('#Percentage of sentences')
ax4.set_ylabel('F1(%)')
ax4.xaxis.set_major_formatter(FuncFormatter(to_percent))


plt.xlabel('training examples')
plt.tight_layout()
plt.show()