Python Data Visualization - The Post-2000 Gaokao Generation

In exam-report coverage, the most effective pieces are often the charts rather than the text. For practicing Python data visualization, Gaokao participation, admissions, and regional distribution are all good example datasets because they naturally fit bar charts, line charts, and maps.

This article walks through a small end-to-end example built around three chart types:

1. Bar charts
2. Line charts
3. A China map

Preparation

1. Libraries

   • charts
   • pyecharts

2. Data

   • Collected directly from Baidu.

Before you start

This article keeps the original older example style, especially for the map section, which matches an older pyecharts API. If you are using a newer local version of pyecharts, some method names or arguments may need small adjustments.

Common Chart Types

Bar charts and line charts are frequently seen and used, so let’s start with the basics.

1. Bar Chart

# Number of Gaokao examinees
gaokao_num = [940,940,...,375]
gaokao_num.reverse()
# Number of admitted students
luqu_num = [700,705,...,221]
luqu_num.reverse()
# Admission rate
luqu_lev= [74.46,75,...,59]
luqu_lev.reverse()

import charts

options = {
    'chart'   : {'zoomType':'xy'},
    # Title
    'title'   : {'text': '2000-2017 Gaokao Data'},
    # Subtitle
    'subtitle': {'text': 'Source: edu.sina.com.cn'},
    # X-axis
    'xAxis'   : {'categories': ['2000',...,'2017']},
    # Y-axis
    'yAxis'   : {'title': {'text': 'Million people/year'}},
    }
series =  [{
    'type': 'column',
    'name': 'Number of Gaokao Examinees',
    'data': gaokao_num
},{
    'type': 'column',
    'name': 'Number of Admitted Students',
    'data': luqu_num
}
]
charts.plot(series, options=options, show='inline')

Due to a minor issue with my pyecharts setup, I used the charts library. Using pyecharts is even simpler, but I won’t repeat it here. You can check the source code if needed.

(Number of Gaokao examinees vs. Number of admitted students)

This year, there are 9.75 million examinees (not visible in the table as I didn’t include it), but you can see that the number of Gaokao examinees remained stable between 2010 and 2018. This is unavoidable given the large population base. The phrase “thousands of soldiers and horses crossing a single-plank bridge” certainly feels apt…

2. Line Chart

If the bar chart above works, then the line chart is very simple; just change the type from column to line.

series =  [{
    'type': 'line',
    'name': 'Admission Rate',
    'data': luqu_lev,
    'marker':{
        'lineWidth':2,
        'lineColor':'#90ed7d'
    }
}
]

Originally, I intended to display both the bar chart and the line chart on a single graph, but the Y-axis values were too large, making the changes in admission rate look insignificant. Therefore, I separated them.

Map Charts

Here’s a Gaokao difficulty ranking. I, too, came from the “hard mode” back in the day. I really feel for the students in Jiangsu; it seems like Ge Jun is designing the math questions again, so Jiangsu will likely be trending again.

Now, let’s use pyecharts to create a map chart and see which province in China has the most Gaokao examinees.

from pyecharts import Map
# Process data
add =["Beijing",...,"Tibet"]
num = [6.3,...,2.53]
dict_data = dict(zip(add,num))
data = [tuple(i) for i in zip(add,num)]
data_pro = data
geo = Map("Distribution of Gaokao Examinees by Province",
          "data from : gaokao.eol.cn",
          title_color="#fff",
          title_pos="center",
          width=1000,
          height=600,
          background_color='#404a59'
          )
attr, value = geo.cast(data_pro)
geo.add("",  attr, value,
          visual_range=[0, 80],
          maptype='china',
          visual_text_color="#fff",
          symbol_size=10,
          is_visualmap=True)
geo.render("各省高考分布.html") # Generates an HTML file
geo # Displays directly in the notebook

Effect Diagram: It seems that students in Henan and Guangdong are under a lot of pressure (!_!)

How to validate the chart output

If you run the code in a notebook, the chart may display inline. In a local script workflow, it is better to verify two things explicitly:

1. The HTML file was created successfully
2. The generated page loads correctly in a browser

For example:

ls 各省高考分布.html

If the file exists but the page is blank, check your pyecharts version first and confirm that the required front-end assets were loaded.

A Little More

I strongly recommend using pyecharts; it’s simple and easy to use. However, for me, only map-related charts work with pyecharts; other types don’t seem to render, so I was forced to use the charts library.

Finally, I hope all Gaokao examinees achieve their desired results so they can celebrate to their heart’s content!

Related reading

• Python: Creating Beautiful Lollipop Charts
• Drawing Raincloud Plots with Python