[数据分析与可视化] 基于plottable库绘制精美表格
plottable是一个Python库，用于在matplotlib中绘制精美定制的图形表格。plottable的官方仓库地址为：plottable。本文主要参考其官方文档，plottable的官方文档地址为：plottable-doc。plottable安装命令如下：
pip install plottable
本文所有代码见：Python-Study-Notes
# jupyter notebook环境去除warning
import warnings
warnings.filterwarnings("ignore")
import plottable
# 打印plottable版本
print('plottable version:', plottable.__version__)
# 打印matplotlib版本
import matplotlib as plt
print('matplotlib version:', plt.__version__)
plottable version: 0.1.5
matplotlib version: 3.5.3目录
1 使用说明
下面的代码展示了一个简单的图形表格绘制示例，plottable提供了Table类以创建和展示图形表格。
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from plottable import Table
# 生成一个包含随机数据的表格
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# 基于pandas表格数据创建和展示图形表格
tab = Table(d)
# 保存图片
plt.savefig("table.jpg", dpi=300,bbox_inches='tight')
plt.show()
对于plottable的Table类，其构造参数介绍如下：
- df: pd.DataFrame, 要显示为表格的DataFrame对象
- ax: mpl.axes.Axes, 绘制表格的坐标轴对象，默认为None
- index_col: str, DataFrame中的索引列名。默认为None
- columns: List[str], 哪些列用于绘图。为None表示使用所有列
- column_definitions: List[ColumnDefinition], 需要设置样式列的style定义类，默认为None
- textprops: Dict[str, Any], 文本属性的字典，默认为空字典
- cell_kw: Dict[str, Any], 单元格属性的字典，默认为空字典
- col_label_cell_kw: Dict[str, Any], 列标签单元格属性的字典，默认为空字典
- col_label_divider: bool, 是否在列标签下方绘制分隔线，默认为True。
- footer_divider: bool, 是否在表格下方绘制分隔线，默认为False。
- row_dividers: bool, 是否显示行分隔线，默认为True
- row_divider_kw: Dict[str, Any], 行分隔线属性的字典，默认为空字典
- col_label_divider_kw: Dict[str, Any], 列标签分隔线属性的字典，默认为空字典
- footer_divider_kw: Dict[str, Any], 页脚分隔线属性的字典，默认为空字典
- column_border_kw: Dict[str, Any], 列边框属性的字典，默认为空字典
- even_row_color: str | Tuple, 偶数行单元格的填充颜色，默认为None
- odd_row_color: str | Tuple, 奇数行单元格的填充颜色，默认为None
在这些参数之中，控制表格绘图效果的参数有以下几类：
- column_definitions：列的样式自定义
- textprops：文本的样样式自定义
- cell_kw：表格单元格的样式自定义
- 其他设置参数的样式
在这些参数中，最重要的参数是column_definitions，因为column_definitions可以控制几乎所有的绘图效果。接下来本文主要对column_definitions的使用进行具体介绍。
plottable提供了ColumnDefinition类（别名ColDef）来自定义图形表格的单个列的样式。ColumnDefinition类的构造参数如下：
name: str，要设置绘图效果的列名
title: str = None，用于覆盖列名的绘图标题
width: float = 1，列的宽度，默认情况下各列的宽度为轴的宽度/列的总数
textprops: Dict[str, Any] = field(default_factory=dict)，提供给每个文本单元格的文本属性
formatter: Callable = None，用于格式化文本外观的可调用函数
cmap: Callable = None，根据单元格的值返回颜色的可调用函数
text_cmap: Callable = None，根据单元格的值返回颜色的可调用函数
group: str = None，设置每个组都会在列标签上方显示的分组列标签
plot_fn: Callable = None，一个可调用函数，将单元格的值作为输入，并在每个单元格上创建一个子图并绘制在其上
要向其传递其他参数
plot_kw: Dict[str, Any] = field(default_factory=dict)，提供给plot_fn的附加关键字参数
border: str | List = None，绘制垂直边界线，可以是"left" / "l"、"right" / "r"或"both"
通过ColumnDefinition类来设置Table类的column_definitions参数，可以实现不同表格列样式的效果。如果是同时多个列的绘图效果，则需要使用[ColumnDefinition,ColumnDefinition]列表的形式。一些使用示例如下
设置列标题和列宽
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from plottable import ColumnDefinition, ColDef, Table
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# name表示设置哪个列的样式
tab = Table(d, column_definitions=[ColumnDefinition(name="A", title="Title A"),
ColumnDefinition(name="D", width=2)])
plt.show()
设置列的文字属性和文本格式
from plottable.formatters import decimal_to_percent
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# index列的文字居中，加粗
# 列A的文本数值改为百分制
tab = Table(d, column_definitions=[ColumnDefinition(name="index", textprops={"ha": "center", "weight": "bold"}),
ColumnDefinition(name="A", formatter=decimal_to_percent)])
plt.show()
设置列单元格背景色和字体颜色
from plottable.cmap import normed_cmap
import matplotlib.cm
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# cmap设置单元格背景色
tab = Table(d, column_definitions=[ColumnDefinition(name="A", cmap=matplotlib.cm.tab20, text_cmap=matplotlib.cm.Reds),
ColumnDefinition(name="B", cmap=matplotlib.cm.tab20b),
ColumnDefinition(name="C", text_cmap=matplotlib.cm.tab20c)])
plt.show()
设置列的分组名
from plottable.cmap import normed_cmap
import matplotlib.cm
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# 将列B和列C视为同一组，该组命名为group_name
tab = Table(d, column_definitions=[ColumnDefinition(name="B", group="group_name"),
ColumnDefinition(name="C", group="group_name")])
plt.show()
设置列边框
from plottable.cmap import normed_cmap
import matplotlib.cm
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# 将列B和列C视为同一组，该组命名为group_name
tab = Table(d, column_definitions=[ColumnDefinition(name="A", border="l"),
ColumnDefinition(name="C", border="both")])
plt.show()
调用函数的使用
ColumnDefinition类的plot_fn和plot_kw参数提供了自定义函数实现表格效果绘制的功能。其中plot_fn表示待调用的函数，plot_kw表示待调用函数的输入参数。此外在plotable.plots预置了一些效果函数，我们可以参考这些效果函数定义自己的绘图函数。预置效果函数如下：
from pathlib import Path
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from matplotlib.colors import LinearSegmentedColormap
from plottable import ColumnDefinition, Table
# 调用预置绘图函数
from plottable.plots import image,monochrome_image,circled_image,bar,percentile_bars,percentile_stars,progress_donut
cmap = matplotlib.cm.tab20
# 柱状图绘制
fig, ax = plt.subplots(figsize=(1, 1))
# 0.7表示数值,lw边框线宽
b = bar(ax, 0.7, plot_bg_bar=True, cmap=cmap, annotate=True, lw=2, height=0.35)
plt.show()
# 星星百分比图
fig, ax = plt.subplots(figsize=(2, 1))
stars = percentile_stars(ax, 85, background_color="#f0f0f0")
# 圆环图
fig, ax = plt.subplots(figsize=(1, 1))
donut = progress_donut(ax, 73, textprops={"fontsize": 14})
plt.show()
对于待调用的函数，可以通过help函数查看这些函数的参数含义。
help(progress_donut)
Help on function progress_donut in module plottable.plots:
progress_donut(ax: matplotlib.axes._axes.Axes, val: float, radius: float = 0.45, color: str = None, background_color: str = None, width: float = 0.05, is_pct: bool = False, textprops: Dict[str, Any] = {}, formatter: Callable = None, **kwargs) -> List[matplotlib.patches.Wedge]
Plots a Progress Donut on the axes.
Args:
ax (matplotlib.axes.Axes): Axes
val (float): value
radius (float, optional):
radius of the progress donut. Defaults to 0.45.
color (str, optional):
color of the progress donut. Defaults to None.
background_color (str, optional):
background_color of the progress donut where the value is not reached. Defaults to None.
width (float, optional):
width of the donut wedge. Defaults to 0.05.
is_pct (bool, optional):
whether the value is given not as a decimal, but as a value between 0 and 100.
Defaults to False.
textprops (Dict[str, Any], optional):
textprops passed to ax.text. Defaults to {}.
formatter (Callable, optional):
a string formatter.
Can either be a string format, ie "{:2f}" for 2 decimal places.
Or a Callable that is applied to the value. Defaults to None.
Returns:
List[matplotlib.patches.Wedge]通过plot_fn和plot_kw参数设置自定义绘图函数和函数输入参数，可以展示不同的绘图效果，如下所示：
from plottable.cmap import normed_cmap
import matplotlib.cm
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# plot_fn和plot_kw
tab = Table(d, textprops={"ha": "center"},
column_definitions=[ColumnDefinition(name="B", plot_fn=bar,plot_kw={'plot_bg_bar':True,'cmap':cmap,
'annotate':True, 'height':0.5}),
ColumnDefinition(name="D", plot_fn=progress_donut,plot_kw={'is_pct':True,})])
plt.show()
自定义文字格式
plottable提供了以下三个自定义函数来表示不同的文字格式：
- decimal_to_percent：将数值数据变为百分比
- tickcross：将数值格式化为或
- signed_integer：添加正负符号
我们可以通过ColumnDefinition的formatter来设置文字格式，如下所示：
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from plottable import ColumnDefinition, Table
from plottable.formatters import decimal_to_percent,tickcross,signed_integer
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
tab = Table(d, column_definitions=[ColumnDefinition(name="A", formatter=decimal_to_percent),
ColumnDefinition(name="C", formatter=tickcross),
ColumnDefinition(name="D", formatter=signed_integer)])
plt.show()
此外，也可以自定义函数来设置文本格式，如下所示：
def setformat(x):
# 使用format格式函数
return "{:.2e}".format(x)
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
tab = Table(d, textprops={"ha": "center"},column_definitions=[ColumnDefinition(name="B", formatter=setformat),
ColumnDefinition(name="D", formatter=lambda x: round(x, 2))])
plt.show()
访问行列单元格
plottable提供了直接访问Table实例的某一行、某一列的方法，如下所示：
from plottable.cmap import normed_cmap
import matplotlib.cm
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# 实例化Table对象
tab = Table(d)
# 根据列名，提取整列
tab.columns['A']
Column(cells=[TextCell(xy=(1, 0), content=0.0, row_idx=0, col_idx=1), TextCell(xy=(1, 1), content=0.09, row_idx=1, col_idx=1), TextCell(xy=(1, 2), content=0.95, row_idx=2, col_idx=1), TextCell(xy=(1, 3), content=0.08, row_idx=3, col_idx=1), TextCell(xy=(1, 4), content=0.92, row_idx=4, col_idx=1)], index=1)
# 读取某列第1行的内容
tab.columns['B'].cells[1].content
0.04
# 根据行索引，提取整行
tab.rows[1]
Row(cells=[TextCell(xy=(0, 1), content=1, row_idx=1, col_idx=0), TextCell(xy=(1, 1), content=0.09, row_idx=1, col_idx=1), TextCell(xy=(2, 1), content=0.04, row_idx=1, col_idx=2), TextCell(xy=(3, 1), content=0.42, row_idx=1, col_idx=3), TextCell(xy=(4, 1), content=0.64, row_idx=1, col_idx=4), TextCell(xy=(5, 1), content=0.26, row_idx=1, col_idx=5)], index=1)
# 提取表头列名
tab.col_label_row
Row(cells=[TextCell(xy=(0, -1), content=index, row_idx=-1, col_idx=0), TextCell(xy=(1, -1), content=A, row_idx=-1, col_idx=1), TextCell(xy=(2, -1), content=B, row_idx=-1, col_idx=2), TextCell(xy=(3, -1), content=C, row_idx=-1, col_idx=3), TextCell(xy=(4, -1), content=D, row_idx=-1, col_idx=4), TextCell(xy=(5, -1), content=E, row_idx=-1, col_idx=5)], index=-1)设置单元格样式
在上面的例子可以看到plottable直接访问表格行列对象，因此我们可以通过设置这些对象的绘图属性来直接更改其绘图效果或文字效果，所支持更改的属性如下：
单元格属性
- set_alpha：设置单元格的透明度。
- set_color：设置单元格的颜色。
- set_edgecolor：设置单元格边缘的颜色。
- set_facecolor：设置单元格内部的颜色。
- set_fill：设置单元格是否填充。
- set_hatch：设置单元格的填充图案。
- set_linestyle：设置单元格边缘线的样式。
- set_linewidth：设置单元格边缘线的宽度。
字体属性
- set_fontcolor：设置字体的颜色。
- set_fontfamily：设置字体的家族。
- set_fontsize：设置字体的大小。
- set_ha：设置文本的水平对齐方式。
- set_ma：设置文本的垂直对齐方式。
示例代码如下：
from plottable.cmap import normed_cmap
import matplotlib.cm
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 5))
# 实例化Table对象
tab = Table(d)
# 设置行号为1的行的背景颜色
tab.rows[1].set_facecolor("grey")
# 设置B列的字体颜色
tab.columns['B'].set_fontcolor("red")
Column(cells=[TextCell(xy=(2, 0), content=0.38, row_idx=0, col_idx=2), TextCell(xy=(2, 1), content=0.69, row_idx=1, col_idx=2), TextCell(xy=(2, 2), content=0.15, row_idx=2, col_idx=2), TextCell(xy=(2, 3), content=0.74, row_idx=3, col_idx=2), TextCell(xy=(2, 4), content=0.41, row_idx=4, col_idx=2)], index=2)
2 绘图实例
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from plottable import Table
d = pd.DataFrame(np.random.random((5, 5)), columns=["A", "B", "C", "D", "E"]).round(2)
fig, ax = plt.subplots(figsize=(6, 3))
# row_dividers显示行的分割线
# odd_row_color奇数行颜色
# even_row_color偶数行颜色
tab = Table(d, row_dividers=False, odd_row_color="#f0f0f0", even_row_color="#e0f6ff")
plt.show()
fig.savefig("table.jpg",dpi=300,bbox_inches='tight')
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from matplotlib.colors import LinearSegmentedColormap
from plottable import ColumnDefinition, Table
from plottable.formatters import decimal_to_percent
from plottable.plots import bar, percentile_bars, percentile_stars, progress_donut
# 自定义颜色
cmap = LinearSegmentedColormap.from_list(
name="BuYl", colors=["#01a6ff", "#eafedb", "#fffdbb", "#ffc834"], N=256
)
fig, ax = plt.subplots(figsize=(6, 6))
d = pd.DataFrame(np.random.random((5, 4)), columns=["A", "B", "C", "D"]).round(2)
tab = Table(
d,
cell_kw={
"linewidth": 0,
"edgecolor": "k",
},
textprops={"ha": "center"},
column_definitions=[
ColumnDefinition("index", textprops={"ha": "left"}),
ColumnDefinition("A", plot_fn=percentile_bars, plot_kw={"is_pct": True}),
ColumnDefinition(
"B", width=1.5, plot_fn=percentile_stars, plot_kw={"is_pct": True}
),
ColumnDefinition(
"C",
plot_fn=progress_donut,
plot_kw={
"is_pct": True,
"formatter": "{:.0%}"
},
),
ColumnDefinition(
"D",
width=1.25,
plot_fn=bar,
plot_kw={
"cmap": cmap,
"plot_bg_bar": True,
"annotate": True,
"height": 0.5,
"lw": 0.5,
"formatter": decimal_to_percent,
},
),
],
)
plt.show()
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from matplotlib.colors import LinearSegmentedColormap
# ColDef是ColumnDefinition的别名
from plottable import ColDef, Table
# 自定义颜色
cmap = LinearSegmentedColormap.from_list(
name="bugw", colors=["#ffffff", "#f2fbd2", "#c9ecb4", "#93d3ab", "#35b0ab"], N=256
)
# 创建数据
cities = [
"TORONTO",
"VANCOUVER",
"HALIFAX",
"CALGARY",
"OTTAWA",
"MONTREAL",
"WINNIPEG",
"EDMONTON",
"LONDON",
"ST. JONES",
]
months = [
"JAN",
"FEB",
"MAR",
"APR",
"MAY",
"JUN",
"JUL",
"AUG",
"SEP",
"OCT",
"NOV",
"DEC",
]
data = np.random.random((10, 12)) + np.abs(np.arange(12) - 5.5)
data = (1 - data / (np.max(data)))
data.shape
(10, 12)
# 绘图
d = pd.DataFrame(data, columns=months, index=cities).round(2)
fig, ax = plt.subplots(figsize=(14, 5))
# 自定义各列的绘图效果
column_definitions = [
ColDef(name, cmap=cmap, formatter=lambda x: "") for name in months
] + [ColDef("index", title="", width=1.5, textprops={"ha": "right"})]
tab = Table(
d,
column_definitions=column_definitions,
row_dividers=False,
col_label_divider=False,
textprops={"ha": "center", "fontname": "Roboto"},
# 设置各个单元格的效果
cell_kw={
"edgecolor": "black",
"linewidth": 0,
},
)
# 设置列标题文字和背景颜色
tab.col_label_row.set_facecolor("white")
tab.col_label_row.set_fontcolor("black")
# 设置行标题文字和背景颜色
tab.columns["index"].set_facecolor("black")
tab.columns["index"].set_fontcolor("white")
tab.columns["index"].set_linewidth(0)
plt.show()
step1 准备数据
下载示例数据，所有示例数据在plottable-example_notebooks。
# 下载数据集
# !wget https://raw.githubusercontent.com/znstrider/plottable/master/docs/example_notebooks/data/wwc_forecasts.csv
from pathlib import Path
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from matplotlib.colors import LinearSegmentedColormap
from plottable import ColumnDefinition, Table
from plottable.cmap import normed_cmap
from plottable.formatters import decimal_to_percent
from plottable.plots import circled_image # image
cols = [
"team",
"points",
"group",
"spi",
"global_o",
"global_d",
"group_1",
"group_2",
"group_3",
"make_round_of_16",
"make_quarters",
"make_semis",
"make_final",
"win_league",
]
# 读取数据
df = pd.read_csv(
"data/wwc_forecasts.csv",
usecols=cols,
)
# 展示数据
df.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
team
group
spi
global_o
global_d
group_1
group_2
group_3
make_round_of_16
make_quarters
make_semis
make_final
win_league
points
0
USA
F
98.32748
5.52561
0.58179
0.82956
0.17044
0.00000
1.0
0.78079
0.47307
0.35076
0.23618
6
1
France
A
96.29671
4.31375
0.52137
0.99483
0.00515
0.00002
1.0
0.78367
0.42052
0.30038
0.19428
6
2
Germany
B
93.76549
3.96791
0.67818
0.98483
0.01517
0.00000
1.0
0.89280
0.48039
0.27710
0.12256
6
3
Canada
E
93.51599
3.67537
0.56980
0.38830
0.61170
0.00000
1.0
0.59192
0.36140
0.20157
0.09031
6
4
England
D
91.92311
3.51560
0.63717
0.70570
0.29430
0.00000
1.0
0.68510
0.43053
0.16465
0.08003
6
此外，我们需要准备每个国家对应的国旗图片，该数据也在plottable-example_notebooks下。
# 读取图片路径
flag_paths = list(Path("data/country_flags").glob("*.png"))
country_to_flagpath = {p.stem: p for p in flag_paths}step2 数据处理
该步需要合并数据，将其转换为plottable可用的数据结构。
# 重置列名
colnames = [
"Team",
"Points",
"Group",
"SPI",
"OFF",
"DEF",
"1st Place",
"2nd Place",
"3rd Place",
"Make Rd Of 16",
"Make Quarters",
"Make Semis",
"Make Finals",
"Win World Cup",
]
col_to_name = dict(zip(cols, colnames))
col_to_name
{'team': 'Team',
'points': 'Points',
'group': 'Group',
'spi': 'SPI',
'global_o': 'OFF',
'global_d': 'DEF',
'group_1': '1st Place',
'group_2': '2nd Place',
'group_3': '3rd Place',
'make_round_of_16': 'Make Rd Of 16',
'make_quarters': 'Make Quarters',
'make_semis': 'Make Semis',
'make_final': 'Make Finals',
'win_league': 'Win World Cup'}
df[["spi", "global_o", "global_d"]] = df[["spi", "global_o", "global_d"]].round(1)
df = df.rename(col_to_name, axis=1)
# 删除Points列
df = df.drop("Points", axis=1)
# 插入列
df.insert(0, "Flag", df["Team"].apply(lambda x: country_to_flagpath.get(x)))
df = df.set_index("Team")
df.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
Flag
Group
SPI
OFF
DEF
1st Place
2nd Place
3rd Place
Make Rd Of 16
Make Quarters
Make Semis
Make Finals
Win World Cup
Team
USA
data/country_flags/USA.png
F
98.3
5.5
0.6
0.82956
0.17044
0.00000
1.0
0.78079
0.47307
0.35076
0.23618
France
data/country_flags/France.png
A
96.3
4.3
0.5
0.99483
0.00515
0.00002
1.0
0.78367
0.42052
0.30038
0.19428
Germany
data/country_flags/Germany.png
B
93.8
4.0
0.7
0.98483
0.01517
0.00000
1.0
0.89280
0.48039
0.27710
0.12256
Canada
data/country_flags/Canada.png
E
93.5
3.7
0.6
0.38830
0.61170
0.00000
1.0
0.59192
0.36140
0.20157
0.09031
England
data/country_flags/England.png
D
91.9
3.5
0.6
0.70570
0.29430
0.00000
1.0
0.68510
0.43053
0.16465
0.08003
step3 绘图
# 设置颜色
cmap = LinearSegmentedColormap.from_list(
name="bugw", colors=["#ffffff", "#f2fbd2", "#c9ecb4", "#93d3ab", "#35b0ab"], N=256
)
team_rating_cols = ["SPI", "OFF", "DEF"]
group_stage_cols = ["1st Place", "2nd Place", "3rd Place"]
knockout_stage_cols = list(df.columns[-5:])
# 单独设置每一列的绘制参数
col_defs = (
# 绘制第一部分效果
[
ColumnDefinition(
name="Flag",
title="",
textprops={"ha": "center"},
width=0.5,
# 设置自定义效果展示函数
plot_fn=circled_image,
),
ColumnDefinition(
name="Team",
textprops={"ha": "left", "weight": "bold"},
width=1.5,
),
ColumnDefinition(
name="Group",
textprops={"ha": "center"},
width=0.75,
),
ColumnDefinition(
name="SPI",
group="Team Rating",
textprops={"ha": "center"},
width=0.75,
),
ColumnDefinition(
name="OFF",
width=0.75,
textprops={
"ha": "center",
# 设置填充方式
"bbox": {"boxstyle": "circle", "pad": 0.35},
},
cmap=normed_cmap(df["OFF"], cmap=matplotlib.cm.PiYG, num_stds=2.5),
group="Team Rating",
),
ColumnDefinition(
name="DEF",
width=0.75,
textprops={
"ha": "center",
"bbox": {"boxstyle": "circle", "pad": 0.35},
},
cmap=normed_cmap(df["DEF"], cmap=matplotlib.cm.PiYG_r, num_stds=2.5),
group="Team Rating",
),
]
# 绘制第二部分效果
+ [
ColumnDefinition(
name=group_stage_cols[0],
title=group_stage_cols[0].replace(" ", "\n", 1),
formatter=decimal_to_percent,
group="Group Stage Chances",
# 设置边框
border="left",
)
]
+ [
ColumnDefinition(
name=col,
title=col.replace(" ", "\n", 1),
formatter=decimal_to_percent,
group="Group Stage Chances",
)
for col in group_stage_cols[1:]
]
# 绘制第三部分效果
+ [
ColumnDefinition(
name=knockout_stage_cols[0],
title=knockout_stage_cols[0].replace(" ", "\n", 1),
formatter=decimal_to_percent,
cmap=cmap,
group="Knockout Stage Chances",
border="left",
)
]
+ [
ColumnDefinition(
name=col,
title=col.replace(" ", "\n", 1),
formatter=decimal_to_percent,
cmap=cmap,
group="Knockout Stage Chances",
)
for col in knockout_stage_cols[1:]
]
)
# 绘图
fig, ax = plt.subplots(figsize=(18, 18))
table = Table(
df,
column_definitions=col_defs,
row_dividers=True,
footer_divider=True,
ax=ax,
textprops={"fontsize": 14},
row_divider_kw={"linewidth": 1, "linestyle": (0, (1, 5))},
col_label_divider_kw={"linewidth": 1, "linestyle": "-"},
column_border_kw={"linewidth": 1, "linestyle": "-"},
).autoset_fontcolors(colnames=["OFF", "DEF"])
plt.show()
# 保存图片
fig.savefig("wwc_table.jpg", facecolor=ax.get_facecolor(), dpi=300,bbox_inches='tight')
step1 准备数据
from pathlib import Path
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from plottable import ColDef, Table
from plottable.plots import image
# 下载联赛数据
# !wget https://projects.fivethirtyeight.com/soccer-api/club/spi_matches.csv
# !wget https://projects.fivethirtyeight.com/soccer-api/club/spi_matches_latest.csv
# 数据地址
FIVETHIRTYEIGHT_URLS = {
"SPI_MATCHES": "https://projects.fivethirtyeight.com/soccer-api/club/spi_matches.csv",
"SPI_MATCHES_LATEST": "https://projects.fivethirtyeight.com/soccer-api/club/spi_matches_latest.csv",
}
# 读取数据
# df = pd.read_csv(FIVETHIRTYEIGHT_URLS["SPI_MATCHES_LATEST"])
df = pd.read_csv("data/spi_matches_latest.csv")
df.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
season
date
league_id
league
team1
team2
spi1
spi2
prob1
prob2
…
importance1
importance2
score1
score2
xg1
xg2
nsxg1
nsxg2
adj_score1
adj_score2
0
2019
2019-03-01
1979
Chinese Super League
Shandong Luneng
Guizhou Renhe
48.22
37.83
0.5755
0.1740
…
45.9
22.1
1.0
0.0
1.39
0.26
2.05
0.54
1.05
0.00
1
2019
2019-03-01
1979
Chinese Super League
Shanghai Greenland
Shanghai SIPG
39.81
60.08
0.2387
0.5203
…
25.6
63.4
0.0
4.0
0.57
2.76
0.80
1.50
0.00
3.26
2
2019
2019-03-01
1979
Chinese Super League
Guangzhou Evergrande
Tianjin Quanujian
65.59
39.99
0.7832
0.0673
…
77.1
28.8
3.0
0.0
0.49
0.45
1.05
0.75
3.15
0.00
3
2019
2019-03-01
1979
Chinese Super League
Wuhan Zall
Beijing Guoan
32.25
54.82
0.2276
0.5226
…
35.8
58.9
0.0
1.0
1.12
0.97
1.51
0.94
0.00
1.05
4
2019
2019-03-01
1979
Chinese Super League
Chongqing Lifan
Guangzhou RF
38.24
40.45
0.4403
0.2932
…
26.2
21.3
2.0
2.0
2.77
3.17
1.05
2.08
2.10
2.10
5 rows × 23 columns
# 筛选德甲联赛数据,并删除为空数据
bl = df.loc[df.league == "German Bundesliga"].dropna()
bl.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
season
date
league_id
league
team1
team2
spi1
spi2
prob1
prob2
…
importance1
importance2
score1
score2
xg1
xg2
nsxg1
nsxg2
adj_score1
adj_score2
497
2022
2022-08-05
1845
German Bundesliga
Eintracht Frankfurt
Bayern Munich
68.47
91.75
0.1350
0.6796
…
32.6
71.9
1.0
6.0
0.83
4.50
0.65
2.72
1.05
5.96
514
2022
2022-08-06
1845
German Bundesliga
VfL Bochum
Mainz
60.73
68.88
0.3568
0.3629
…
33.5
34.5
1.0
2.0
1.00
1.62
0.96
0.86
1.05
2.10
515
2022
2022-08-06
1845
German Bundesliga
Borussia Monchengladbach
TSG Hoffenheim
69.38
66.77
0.4872
0.2742
…
40.2
33.3
3.0
1.0
1.86
0.10
2.51
0.31
2.36
1.05
516
2022
2022-08-06
1845
German Bundesliga
VfL Wolfsburg
Werder Bremen
68.18
59.82
0.5319
0.2014
…
30.2
33.3
2.0
2.0
0.81
0.97
1.07
1.25
2.10
2.10
517
2022
2022-08-06
1845
German Bundesliga
1. FC Union Berlin
Hertha Berlin
69.98
59.70
0.5479
0.1860
…
34.9
33.0
3.0
1.0
1.25
0.40
1.66
0.36
3.15
1.05
5 rows × 23 columns
step2 数据处理
# 统计得分
def add_points(df: pd.DataFrame) -> pd.DataFrame:
# 三元表达式
# df["score1"] > df["score2"]，则返回3
# np.where(df["score1"] == df["score2"]，则返回1
# 否则为0
df["pts_home"] = np.where(
df["score1"] > df["score2"], 3, np.where(df["score1"] == df["score2"], 1, 0)
)
df["pts_away"] = np.where(
df["score1"] < df["score2"], 3, np.where(df["score1"] == df["score2"], 1, 0)
)
return df
# 统计得分数据
bl = add_points(bl)
# 总得分、总进球数、总助攻数和总黄牌数
# 以下代码先分别统计team1和team2的得分数据，然后将两组数据相加
perform = (
bl.groupby("team1")[[
"pts_home",
"score1",
"score2",
"xg1",
"xg2",
]]
.sum()
.set_axis(
[
"pts",
"gf",
"ga",
"xgf",
"xga",
],
axis=1,
)
.add(
bl.groupby("team2")[[
"pts_away",
"score2",
"score1",
"xg2",
"xg1",
]]
.sum()
.set_axis(
[
"pts",
"gf",
"ga",
"xgf",
"xga",
],
axis=1,
)
)
)
# 由于python和pandas版本问题，如果上面的代码出问题，则使用下面代码
# t1= bl.groupby("team1")[["pts_home","score1","score2","xg1","xg2", ]]
# t1 = t1.sum()
# t1.set_axis( ["pts","gf","ga","xgf","xga", ], axis=1,)
# t2 = bl.groupby("team1")[["pts_home","score1","score2","xg1","xg2", ]]
# t2 = t2.sum()
# t2.set_axis( ["pts","gf","ga","xgf","xga", ], axis=1,)
# perform = (t1.add(t2))
perform.shape
(18, 5)
# 汇总得分表现数据
perform.index.name = "team"
perform["gd"] = perform["gf"] - perform["ga"]
perform = perform[
[
"pts",
"gd",
"gf",
"ga",
"xgf",
"xga",
]
]
perform["games"] = bl.groupby("team1").size().add(bl.groupby("team2").size())
perform.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
pts
gd
gf
ga
xgf
xga
games
team
1. FC Union Berlin
62
13.0
51.0
38.0
35.93
43.06
34
Bayer Leverkusen
50
8.0
57.0
49.0
53.62
48.20
34
Bayern Munich
71
54.0
92.0
38.0
84.93
40.12
34
Borussia Dortmund
71
39.0
83.0
44.0
75.96
42.69
34
Borussia Monchengladbach
43
-3.0
52.0
55.0
53.05
58.88
34
# 统计各队的胜负数据
def get_wins_draws_losses(games: pd.DataFrame) -> pd.DataFrame:
return (
games.rename({"pts_home": "pts", "team1": "team"}, axis=1)
.groupby("team")["pts"]
.value_counts()
.add(
games.rename({"pts_away": "pts", "team2": "team"}, axis=1)
.groupby("team")["pts"]
.value_counts(),
fill_value=0,
)
.astype(int)
.rename("count")
.reset_index(level=1)
.pivot(columns="pts", values="count")
.rename({0: "L", 1: "D", 3: "W"}, axis=1)[["W", "D", "L"]]
)
wins_draws_losses = get_wins_draws_losses(bl)
wins_draws_losses.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
pts
W
D
L
team
1. FC Union Berlin
18
8
8
Bayer Leverkusen
14
8
12
Bayern Munich
21
8
5
Borussia Dortmund
22
5
7
Borussia Monchengladbach
11
10
13
# 合并得分和胜负数据
perform = pd.concat([perform, wins_draws_losses], axis=1)
perform.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
pts
gd
gf
ga
xgf
xga
games
W
D
L
team
1. FC Union Berlin
62
13.0
51.0
38.0
35.93
43.06
34
18
8
8
Bayer Leverkusen
50
8.0
57.0
49.0
53.62
48.20
34
14
8
12
Bayern Munich
71
54.0
92.0
38.0
84.93
40.12
34
21
8
5
Borussia Dortmund
71
39.0
83.0
44.0
75.96
42.69
34
22
5
7
Borussia Monchengladbach
43
-3.0
52.0
55.0
53.05
58.88
34
11
10
13
step3 映射队标图片
队标图片地址为：plottable-example_notebooks
# 创建队名和队标的索引数据
club_logo_path = Path("data/bundesliga_crests_22_23")
club_logo_files = list(club_logo_path.glob("*.png"))
club_logos_paths = {f.stem: f for f in club_logo_files}
perform = perform.reset_index()
# 添加新列
perform.insert(0, "crest", perform["team"])
perform["crest"] = perform["crest"].replace(club_logos_paths)
# 数据排序
perform = perform.sort_values(by=["pts", "gd", "gf"], ascending=False)
for colname in ["gd", "gf", "ga"]:
perform[colname] = perform[colname].astype("int32")
perform["goal_difference"] = perform["gf"].astype(str) + ":" + perform["ga"].astype(str)
# 添加排名
perform["rank"] = list(range(1, 19))
perform.head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
crest
team
pts
gd
gf
ga
xgf
xga
games
W
D
L
goal_difference
rank
2
data/bundesliga_crests_22_23/Bayern Munich.png
Bayern Munich
71
54
92
38
84.93
40.12
34
21
8
5
92:38
1
3
data/bundesliga_crests_22_23/Borussia Dortmund…
Borussia Dortmund
71
39
83
44
75.96
42.69
34
22
5
7
83:44
2
10
data/bundesliga_crests_22_23/RB Leipzig.png
RB Leipzig
66
23
64
41
67.01
37.48
34
20
6
8
64:41
3
0
data/bundesliga_crests_22_23/1. FC Union Berli…
1. FC Union Berlin
62
13
51
38
35.93
43.06
34
18
8
8
51:38
4
11
data/bundesliga_crests_22_23/SC Freiburg.png
SC Freiburg
59
7
51
44
53.11
52.25
34
17
8
9
51:44
5
step4 设定绘图数据
# 设置颜色
row_colors = {
"top4": "#2d3636",
"top6": "#516362",
"playoffs": "#8d9386",
"relegation": "#c8ab8d",
"even": "#627979",
"odd": "#68817e",
}
bg_color = row_colors["odd"]
text_color = "#e0e8df"
# 确定绘图列
table_cols = ["crest", "team", "games", "W", "D", "L", "goal_difference", "gd", "pts"]
perform[table_cols].head().dataframe tbody tr th:only-of-type { vertical-align: middle }
\3cpre>\3ccode>.dataframe tbody tr th { vertical-align: top }
.dataframe thead th { text-align: right }
crest
team
games
W
D
L
goal_difference
gd
pts
2
data/bundesliga_crests_22_23/Bayern Munich.png
Bayern Munich
34
21
8
5
92:38
54
71
3
data/bundesliga_crests_22_23/Borussia Dortmund…
Borussia Dortmund
34
22
5
7
83:44
39
71
10
data/bundesliga_crests_22_23/RB Leipzig.png
RB Leipzig
34
20
6
8
64:41
23
66
0
data/bundesliga_crests_22_23/1. FC Union Berli…
1. FC Union Berlin
34
18
8
8
51:38
13
62
11
data/bundesliga_crests_22_23/SC Freiburg.png
SC Freiburg
34
17
8
9
51:44
7
59
# 定义各列绘图效果
table_col_defs = [
ColDef("rank", width=0.5, title=""),
ColDef("crest", width=0.35, plot_fn=image, title=""),
ColDef("team", width=2.5, title="", textprops={"ha": "left"}),
ColDef("games", width=0.5, title="Games"),
ColDef("W", width=0.5),
ColDef("D", width=0.5),
ColDef("L", width=0.5),
ColDef("goal_difference", title="Goals"),
ColDef("gd", width=0.5, title="", formatter="{:+}"),
ColDef("pts", border="left", title="Points"),
]step5 绘图
fig, ax = plt.subplots(figsize=(14, 12))
plt.rcParams["text.color"] = text_color
# 设置绘图字体
# plt.rcParams["font.family"] = "Roboto"
# 设置背景颜色
fig.set_facecolor(bg_color)
ax.set_facecolor(bg_color)
table = Table(
perform,
column_definitions=table_col_defs,
row_dividers=True,
col_label_divider=False,
footer_divider=True,
index_col="rank",
columns=table_cols,
even_row_color=row_colors["even"],
footer_divider_kw={"color": bg_color, "lw": 2},
row_divider_kw={"color": bg_color, "lw": 2},
column_border_kw={"color": bg_color, "lw": 2},
# 如果设置字体需要添加"fontname": "Roboto"
textprops={"fontsize": 16, "ha": "center"},
)
# 设置不同行的颜色
for idx in [0, 1, 2, 3]:
table.rows[idx].set_facecolor(row_colors["top4"])
for idx in [4, 5]:
table.rows[idx].set_facecolor(row_colors["top6"])
table.rows[15].set_facecolor(row_colors["playoffs"])
for idx in [16, 17]:
table.rows[idx].set_facecolor(row_colors["relegation"])
table.rows[idx].set_fontcolor(row_colors["top4"])
fig.savefig(
"bohndesliga_table_recreation.png",
facecolor=fig.get_facecolor(),
bbox_inches='tight',
dpi=300,
)
3 参考
手机扫一扫
移动阅读更方便
你可能感兴趣的文章