用Python把图做得好看点：用Matplotlib画个Circos和弦图

此文于 2019-02-23 发表在EasyCharts-知乎专栏：Python数据分析与可视化

https://zhuanlan.zhihu.com/EasyCharts-Python

这系列文章就是下面这位身高180、帅气的小哥哥撰写，人长得好看，画的图也好看！

一直觉得circos图很好看，最近尝试了一下对中间关系曲线的绘制

主要利用的数学工具是贝塞尔曲线

但是对贝塞尔曲线中间的控制点取值还是不太满意造成部分曲线看着有点奇怪

成品大概是这样

代码如下：

import matplotlib.pyplot as pltimport numpy as npimport pandas as pdimport random
Start1=[0 for i in range(50)]for i in range(60):    Start1.append(5/np.power(2,0.2))for i in range(80):    Start1.append(-5/np.power(2,0.2))
df=pd.DataFrame(Start1,columns=['Startx'])df['Starty']=np.power(25-np.power(df['Startx'],2),0.5)*-1df['Endx']=[random.uniform(-5,5.0) for i in range(len(df))]  df['Endy']=np.power(25-np.power(df['Endx'],2),0.5)*[random.sample([1,-1], 1)[0] for i in range(len(df))]df['']=np.power(25


    
-np.power(df['Endx'],2),0.5)df['斜率1']=(df['Starty']-df['Endy'])/(df['Startx']-df['Endx'])df['斜率2']=-1/df['斜率1']df['中心点x']=(df['Startx']+df['Endx'])/2df['中心点y']=(df['Starty']+df['Endy'])/2df['x轴移动']=df['中心点x']-df['中心点x']*(0.05)df['y轴移动']=df['x轴移动']*df['斜率2']df['控制点x']=df['中心点x']-df['x轴移动']df['控制点y']=df['中心点y']-df['y轴移动']df['r1']=np.power(df['Startx']*df['Startx']+df['Starty']*df['Starty'],1/2)df['theta1']=np.arctan2(df['Starty'],df['Startx'])df['r2']=np.power(df['Endx']*df['Endx']+df['Endy']*df['Endy'],1/2)df['theta2']=np.arctan2(df['Endy'],df['Endx'])fig=plt.figure(figsize=[15,15])ax=fig.add_subplot(1,1,1,projection='polar')t=np.linspace(0,1,1000)for i in range(len(df)):    x_c=(1-t)*(1-t)*df.iloc[i]['Startx']+2*t*(1-t)*df.iloc[i]['控制点x']+t*t*df.iloc[i]['Endx']    y_c=(1-t)*(1-t)*df.iloc[i]['Starty']+2*t*(1-t)*df.iloc[i]['控制点y']+t*t*df.iloc[i]['Endy']    r=np.power(y_c*y_c+x_c*x_c,0.5)    theta=np.arctan2(y_c,x_c)    ax.plot(theta,r,color=(0/255,123/255,187/255),alpha=0.5)ax.scatter(df[['theta1','theta2']],df[['r1','r2']],color=(0/255,123/255,87/255),alpha=1)ax.bar(x=df['theta2'],height=[2],bottom=6,width=0.01,color=(225/255,225/255,225/255),alpha=0.5)plt.axis('off')