matplotlib 图表 - 创建水平条形图

Question

我偶然发现了以下 sn-p，用于使用 matplotlib 创建水平条形图：

import matplotlib
from pylab import *

val = 3+10*rand(5)    # the bar lengths
pos = arange(5)+.5    # the bar centers on the y axis
print pos
figure(1)
barh(pos,val, align='center')
yticks(pos, ('Tom', 'Dick', 'Harry', 'Slim', 'Jim'))
xlabel('Performance')
title('horizontal bar chart using matplotlib')
grid(True)
show()

我想将上面的脚本修改如下：

1. 使绘制的条“不那么粗”（即降低绘制的水平条的高度）
2. 在同一图上将负数和正数绘制为水平条

任何帮助我进行上述修改的帮助（代码 sn-p 或链接）都会非常有帮助。

顺便说一句，如果我想制作堆叠的水平条（假设每个标签有 3 个堆叠的水平条），我将如何修改上面的代码来绘制 3 个堆叠的水平条图？

[[编辑]]

有人可以发布两个短代码 sn-p 说明如何：

1. 在水平条的另一侧打印标签（例如，“负”条的标签出现在第一象限，“正”条的标签出现在第二象限

2. 绘制多个（比如 2 或 3 个）水平条（而不仅仅是一个）。很好的例子是first two images shown here

Answer 1

下面的代码 sn-p 是一个例子，使用 text 函数在左侧标注负值，右侧标注正值，gcalmettes 和zhenya 都提到过。

from pylab import setp
import numpy as np
import matplotlib.pyplot as plt
import math

# creation of the data
name_list = ['day1', 'day2', 'day3', 'day4']
data = {name: 3+10*np.random.rand(5) for name in name_list}

for name in name_list:
  data[name][0] = data[name][0]*-1
  data[name][2] = data[name][2]*-1

colors_list = ['0.5', 'r', 'b', 'g'] #optional

def customize_barh(data, width_bar=1, width_space=0.5, colors=None):
    n_measure = len(data)                   #number of measure per people
    n_people = data[data.keys()[0]].size    # number of people

    #some calculation to determine the position of Y ticks labels
    total_space = n_people*(n_measure*width_bar)+(n_people-1)*width_space
    ind_space = n_measure*width_bar
    step = ind_space/2.
    pos = np.arange(step, total_space+width_space, ind_space+width_space)
    # create the figure and the axes to plot the data 
    fig = plt.figure(figsize=(8,6))
    ax = fig.add_axes([0.15, 0.15, 0.65, 0.7])

    # remove top and right spines and turn ticks off if no spine
    ax.spines['right'].set_color('none')
    ax.spines['top'].set_color('none')
    ax.xaxis.set_ticks_position('bottom')
    ax.yaxis.set_ticks_position('default')    # ticks position on the right
    # postition of tick out
    ax.tick_params(axis='both', direction='out', width=3, length=6,
                   labelsize=24, pad=8)
    ax.spines['left'].set_linewidth(3)
    ax.spines['bottom'].set_linewidth(3)

    # plot the data
    for i,day in enumerate(data.keys()):
        if colors == None:
            ax.barh(pos-step+i*width_bar, data[day], width_bar, #facecolor='0.4',
                    edgecolor='k', linewidth=3)
        else:
            ax.barh(pos-step+i*width_bar, data[day], width_bar, facecolor=colors[i],
                    edgecolor='k', linewidth=3)


    ax.set_yticks(pos)
    # you may want to use the list of name as argument of the function to be more
    # flexible (if you have to add a people)
    setp(ax.get_yticklabels(), visible=False)         
    ax.set_ylim((-width_space, total_space+width_space))
    ax.set_xlabel('Performance', size=26, labelpad=10)
    labels_list = ['Tom', 'Dick', 'Harry', 'Slim','Jim']

    # creation of an array of positive/negative values (based on the values
    # of the data) that will be used as x values for adding text as side labels
    side_list = []
    for index in range(len(labels_list)):
        sum = 0
        for name in name_list:
            sum+= data[name][index]
        if math.copysign(1,sum) > 0:
            side_list.append(16)
        else:
            side_list.append(-21)
    for label in labels_list:
        plt.text(side_list[labels_list.index(label)], pos[labels_list.index(label)]-0.5, label,fontsize=26) 
customize_barh(data, colors=colors_list)
plt.savefig('perf.png')
plt.show()

它的工作原理是，给定人的所有条都需要是负数或正数，才能在正确的一侧注释文本。要更改此行为，只需更改 side_list 的生成即可。

例如，如果您想要某个条形阈值来确定标签的位置，则计算超过该阈值的数据值，而不是对给定名称的值求和。

例如，对于 3 条的阈值，无论多少条，for 循环都变为

for index in range(len(labels_list)):
        count = 0
            for name in name_list:
               if data[name][index] > 0:
                  count+= 1
            if count > 3:
              side_list.append(16)
            else:
              side_list.append(-21)

side_list 的生成也需要更改以适应您的数据范围，因为给出的示例使用指定范围内的随机数据。

例如，您需要调整 side_list.append(16) 和 side_list.append(-21) 的标签偏移量以适合您的数据。

Answer 2

正如珍亚所说，你将不得不调整你的情节。

例如，下面是一个生成自定义水平条形图的函数：

• 输入是数据，包含在字典中
• 然后它会根据您在每个类别（人）中的度量（条）数量以及您希望在每个类别之间放置的空间来计算 Y 刻度的位置。
• 最后它会绘制每个数据度量（如果您已指定，则使用不同的颜色）

默认情况下，它会将类别（人）的名称绘制在右侧，但您当然可以更改它。

import numpy as np
import matplotlib.pyplot as plt

# creation of the data
name_list = ['day1', 'day2', 'day3', 'day4']
data = {name: 3+10*np.random.rand(5) for name in name_list}

colors_list = ['0.5', 'r', 'b', 'g'] #optional

def customize_barh(data, width_bar=1, width_space=0.5, colors=None):
    n_measure = len(data)                   #number of measure per people
    n_people = data[data.keys()[0]].size    # number of people

    #some calculation to determine the position of Y ticks labels
    total_space = n_people*(n_measure*width_bar)+(n_people-1)*width_space
    ind_space = n_measure*width_bar
    step = ind_space/2.
    pos = np.arange(step, total_space+width_space, ind_space+width_space)

    # create the figure and the axes to plot the data 
    fig = plt.figure(figsize=(8,6))
    ax = fig.add_axes([0.15, 0.15, 0.65, 0.7])

    # remove top and right spines and turn ticks off if no spine
    ax.spines['right'].set_color('none')
    ax.spines['top'].set_color('none')
    ax.xaxis.set_ticks_position('bottom')
    ax.yaxis.set_ticks_position('right')    # ticks position on the right
    # postition of tick out
    ax.tick_params(axis='both', direction='out', width=3, length=6,
                   labelsize=24, pad=8)
    ax.spines['left'].set_linewidth(3)
    ax.spines['bottom'].set_linewidth(3)

    # plot the data
    for i,day in enumerate(data.keys()):
        if colors == None:
            ax.barh(pos-step+i*width_bar, data[day], width_bar, #facecolor='0.4',
                    edgecolor='k', linewidth=3)
        else:
            ax.barh(pos-step+i*width_bar, data[day], width_bar, facecolor=colors[i],
                    edgecolor='k', linewidth=3)


    ax.set_yticks(pos)
    # you may want to use the list of name as argument of the function to be more
    # flexible (if you have to add a people)
    ax.set_yticklabels(('Tom', 'Dick', 'Harry', 'Slim', 'Jim'))         
    ax.set_ylim((-width_space, total_space+width_space))
    ax.set_xlabel('Performance', size=26, labelpad=10)

customize_barh(data, colors=colors_list)
plt.savefig('perf.png')
plt.show()

产生：

Answer 3

import matplotlib
from pylab import *

val = 3-6*rand(5)    # the bar lengths        # changed your data slightly
pos = arange(5)+.5    # the bar centers on the y axis
print pos
figure(1)
barh(pos,val, align='center',height=0.1)    # notice the 'height' argument
yticks(pos, ('Tom', 'Dick', 'Harry', 'Slim', 'Jim'))

gca().axvline(0,color='k',lw=3)   # poor man's zero level

xlabel('Performance')
title('horizontal bar chart using matplotlib')
grid(True)
show()

一般来说，我建议不要使用from pyplot import *。除非您处于交互模式，否则请使用面向对象的方法：

import matplotlib.pyplot as plt
from numpy.random import rand
from numpy import arange

val = 3-6*rand(5)    # the bar lengths
pos = arange(5)+.5    # the bar centers on the y axis
print pos

fig = plt.figure()
ax = fig.add_subplot(111)
ax.barh(pos,val, align='center',height=0.1)
ax.set_yticks(pos, ('Tom', 'Dick', 'Harry', 'Slim', 'Jim'))

ax.axvline(0,color='k',lw=3)   # poor man's zero level

ax.set_xlabel('Performance')
ax.set_title('horizontal bar chart using matplotlib')
ax.grid(True)
plt.show()

各种情节的一个很好的起点是matplotlib gallery