## Python Seaborn economist's classic chart imitation

DataCharm 2021-02-22 23:07:44
python seaborn economist classic chart

The last original tweet used R-ggplot2 The realization of the economist's classic chart imitation R-ggplot2 The classic economist's chart imitates , So what about this issue , We just use Python-seaborn To reproduce this classic economist chart . The main knowledge points involved are as follows ：

• Python-seaborn regplot Regression linear fitting graph drawing
• matplotlib Customized drawing of drawing legend
• adjustText Library to avoid adding text

## Python-seaborn Draw a fitting line graph

First , Let's preview the data ( part )：

among Region_new Columns are new columns that are changed according to relevant requirements , Mapping is also based on secondary data .

Use seaborn By drawing the fitting line, you can avoid making your own wheels , Now let's go straight to the basics ( Without any embellishment ), The code is as follows ：

```fig,ax = plt.subplots(figsize=(8,4.5),dpi=200,facecolor='white',edgecolor='white')
ax.set_facecolor("white")
fit_line = sns.regplot(data=test_data,x="CPI",y="HDI",logx=True,ax=ax)
ax.text(.85,-.07,'\nVisualization by DataCharm',transform = ax.transAxes,
ha='center', va='center',fontsize = 8,color='black')
```

The visualization is as follows ：

The main parameters needed here are as follows ：

1. logx ： Used to plot logarithmic fit curves , The default is False, That is to draw a linear fitting line .
2. ci ： That is to draw the confidence interval of fitting curve , It can be (0~100) The integer of , Can also be set to False, That is, no confidence interval is drawn .
3. {scatter,line}_kws ： Dictionary type , You can customize the drawing properties of points and lines , Including color 、 size 、 The thickness etc.

At present, only these ( Because drawing needs ), For more details, please refer to the corresponding official website ：seaborn.regplot

Let's just put in the visual code that we're drawing , And then explain it separately , The code is as follows ：

```fig,ax = plt.subplots(figsize=(8,4.5),dpi=200,facecolor='white',edgecolor='white')
ax.set_facecolor("white")
color = [region_color[i] for i in test_data['Region_new']]
fit_line = sns.regplot(data=test_data,x="CPI",y="HDI",logx=True,ci=False,
line_kws={"color":"red",
"label":r"\$R^2\$=56%",
"lw":1.5},
scatter_kws={"s":50,"fc":"white",
"ec":color,
"lw":1.5,
"alpha":1},
ax=ax)
texts = []
for i, j ,t in zip(data_text["CPI"],data_text["HDI"],data_text["Country"]):
texts.append(ax.annotate(t,xy=(i, j),xytext=(i-.8,j),
arrowprops=dict(arrowstyle="-", color="black",lw=.5),
color='black',size=9))
adjust_text(texts,only_move={'text': 'xy','objects':'x','point':'y'})
#adjust_text(texts,only_move={'text': 'xy'})
ax.set_xlabel("Corruption Perceptions Index, 2011 (10=least corrupt)",fontstyle="italic",
fontsize=8)
ax.set_ylabel("Human Development Index, 2011 (1=best)",fontstyle="italic",fontsize=8)
ax.set_xlim((.5, 10.2))
ax.set_ylim((.2, 1))
ax.set_xticks(np.arange(1, 10.3, step=1))
ax.set_yticks(np.arange(0.2, 1.05, step=0.1))
# Grid settings
ax.grid(which='major',axis='y',ls='-',c='gray',)
ax.set_axisbelow(True)
# Axis ridge setting
for spine in ['top','left','right']:
ax.spines[spine].set_visible(None) # Remove the axial ridge
ax.spines['bottom'].set_color('k') # Set up bottom Color
# Scale setting , Display only bottom The scale of , And the direction is outward , Long 、 Width is also set
ax.tick_params(bottom=True,direction='in',labelsize=12,width=1,length=3,
left=False)
# Add legend
ax.scatter([], [], ec='#01344A', fc="white",label='OECD', lw=1.5)
ax.scatter([], [], ec='#228DBD', fc="white",label='Americas', lw=1.5)
ax.scatter([], [], ec='#6DBBD8', fc="white",label='Asia & \nOceania', lw=1.5)
ax.scatter([], [], ec='#1B6E64', fc="white",label='Central & \nEastern Europe', lw=1.5)
ax.scatter([], [], ec='#D24131', fc="white",label='Middle East & \nnorth Africa', lw=1.5)
ax.scatter([], [], ec='#621107', fc="white",label='Sub-Saharan \nAfrica', lw=1.5)
ax.legend(loc="upper center",frameon=False,ncol=7,fontsize=6.5,bbox_to_anchor=(0.5, 1.1))
ax.text(.5,1.19,"Corruption and human development",transform = ax.transAxes,ha='center',
va='center',fontweight="bold",fontsize=16)
ax.text(.5,1.12, "Base Charts:Scatter Exercise in Python",
transform = ax.transAxes,ha='center', va='center',fontsize = 12,color='black')
ax.text(.9,.05,'\nVisualization by DataCharm',transform = ax.transAxes,
ha='center', va='center',fontsize = 8,color='black')
```

「 Knowledge point 」

• Construction of color dictionary , Easy to assign scatter colors
```color = ('#01344A','#228DBD','#6DBBD8','#1B6E64','#D24131','#621107')
region =("OECD","Americas","Asia & \nOceania","Central & \nEastern Europe",
"Middle East & \nnorth Africa", "Sub-Saharan \nAfrica")
region_color = dict(zip(region,color))
color = [region_color[i] for i in test_data['Region_new']]
# stay regplot() Call the following
scatter_kws={"s":50,"fc":"white",
"ec":color,
"lw":1.5,
"alpha":1}
```
• adjust_text() Method add ax.annotate attribute
```texts = []
for i, j ,t in zip(data_text["CPI"],data_text["HDI"],data_text["Country"]):
texts.append(ax.annotate(t,xy=(i, j),xytext=(i-.8,j),
arrowprops=dict(arrowstyle="-", color="black",lw=.5),
color='black',size=9))
adjust_text(texts,only_move={'text': 'xy','objects':'x','point':'y'})
```
• matplotlib Customized legend settings
```# Add legend
ax.scatter([], [], ec='#01344A', fc="white",label='OECD', lw=1.5)
ax.scatter([], [], ec='#228DBD', fc="white",label='Americas', lw=1.5)
ax.scatter([], [], ec='#6DBBD8', fc="white",label='Asia & \nOceania', lw=1.5)
ax.scatter([], [], ec='#1B6E64', fc="white",label='Central & \nEastern Europe', lw=1.5)
ax.scatter([], [], ec='#D24131', fc="white",label='Middle East & \nnorth Africa', lw=1.5)
ax.scatter([], [], ec='#621107', fc="white",label='Sub-Saharan \nAfrica', lw=1.5)
ax.legend(loc="upper center",frameon=False,ncol=7,fontsize=6.5,bbox_to_anchor=(0.5, 1.1))
```

The final visualization is as follows ：

## summary

In this issue, we've launched Python-seaborn The classic visualization of tweets , Although there are still problems in the final result ( Of course , You can customize the specific location to solve ), But its main purpose is to let you learn drawing skills , Especially the drawing of fitting curve ( If you have wheels, you can use them directly , Don't think about reinventing yourself ).

This article is from WeChat official account. - DataCharm（shujumeili） , author ： Ning Haitao

The source and reprint of the original text are detailed in the text , If there is any infringement , Please contact the yunjia_community@tencent.com Delete .

Original publication time ： 2020-11-22

Participation of this paper Tencent cloud media sharing plan , You are welcome to join us , share .

https://pythonmana.com/2021/02/20210222130146246A.html