Python Matplotlib数据可视化 绘制箱形图、散点图和直方图

import pandas as pd
import matplotlib.pyplot as plt
import matplotlib as mpl

df = pd.read_csv('soccer.csv', encoding='gbk')
labels = [f'等级{i}' for i in ['一', '二', '三', '四', '五']]

data1 = df[df['Skill_Moves'] == 1]['Rating']
data2 = df[df['Skill_Moves'] == 2]['Rating']
data3 = df[df['Skill_Moves'] == 3]['Rating']
data4 = df[df['Skill_Moves'] == 4]['Rating']
data5 = df[df['Skill_Moves'] == 5]['Rating']

# 设置中文显示
mpl.rcParams['font.family'] = 'SimHei'
# 设置图形显示风格
plt.style.use('ggplot')
fig, ax = plt.subplots()
box_plot = ax.boxplot((data1, data2, data3, data4, data5), labels=labels,
            boxprops={'color': 'black'}, showmeans=True, patch_artist=True,
            )

colors = ['pink', 'blue', 'green', 'yellow', 'red']

# 填充箱子颜色
for patch, color in zip(box_plot['boxes'], colors):
    patch.set(facecolor=color)

# 设置箱子两端线的属性
for whisker in box_plot['whiskers']:
    whisker.set(color='purple', linewidth=2)
# 设置顶端和末端线条的属性
for cap in box_plot['caps']:
    cap.set(color='g', linewidth=3)
# 设置中位数的属性
for median in box_plot['medians']:
    median.set(color='black', linewidth=3)

plt.xlabel('技术等级')
plt.ylabel('评分')
plt.title('不同技术等级的运动员评分分布箱形图')

plt.show()

import pandas as pd
import matplotlib.pyplot as plt
import matplotlib as mpl

df= pd.read_csv('soccer.csv', encoding='gbk')
age, rating = list(df['Age']), list(df['Rating'])

# 设置中文显示
mpl.rcParams['font.family'] = 'SimHei'
# 设置图形显示风格
plt.style.use('ggplot')
# 设置大小  像素
plt.figure(figsize=(9, 6), dpi=100)

# 绘制散点图
plt.scatter(age, rating)

# 添加描述信息
plt.title('运动员年龄与评分散点图')
plt.xlabel('年龄')
plt.ylabel('评分')
plt.show()

import pandas as pd
import matplotlib.pyplot as plt
import matplotlib as mpl

ages = list(pd.read_csv('soccer.csv', encoding='gbk')['Age'])
ages.sort()

# 设置中文显示
mpl.rcParams['font.family'] = 'SimHei'
# 设置图形显示风格
plt.style.use('ggplot')
plt.figure(figsize=(9, 6), dpi=100)

bin_width = 1  # 设置组距   整除
num_bin = (max(ages) - min(ages)) // bin_width    # 组数

# 绘制直方图  x：指定要绘制直方图的数据
# bins：指定直方图条形的个数  color：设置直方图的填充色    edgecolor：指定直方图的边界色
plt.hist(x=ages, bins=num_bin, color='blue', edgecolor='k', label='直方图')   # 为直方图呈现标签
plt.xticks(range(20, 50, 5))     # 设置x轴刻度

# 添加描述信息
plt.xlabel('年龄区间')
plt.ylabel('频数')
plt.title('年龄频数分布直方图')

plt.legend()
plt.show()