如何订购数据框以在熊猫中绘制 3d 条形图答案

【问题标题】：How to order dataframe for plotting 3d bar in pandas如何订购数据框以在熊猫中绘制 3d 条形图
【发布时间】：2014-05-21 03:48:51
【问题描述】：

我正在尝试从 pandas 创建一个包含多个 3d 条形的图表。查看网络上的一些示例，我发现完成此操作的最佳方法是获取这样的数据框：

data

Variable         A         B          C          D
date
2000-01-03  0.469112  -1.135632   0.119209   -2.104569
2000-01-04 -0.282863   1.212112  -1.044236   -0.494929
2000-01-05 -1.509059  -0.173215  -0.861849    1.071804

我的数据框是：

df

       Date_inicio             Date_Fin             Date_Max       Clase
0   2004-04-09 23:00:00  2004-04-10 04:00:00  2004-04-10 02:00:00  MBCCM
1   2004-04-12 23:00:00  2004-04-13 04:00:00  2004-04-13 00:00:00  MBSCL
2   2004-04-24 04:00:00  2004-04-24 12:00:00  2004-04-24 09:00:00    SCL
3   2004-05-02 07:00:00  2004-05-02 14:00:00  2004-05-02 11:00:00    SCL
4   2004-05-30 05:00:00  2004-05-30 08:00:00  2004-05-30 07:00:00  MBCCM
5   2004-05-31 03:00:00  2004-05-31 07:00:00  2004-05-31 05:00:00  MBCCM   
6   2004-06-08 00:00:00  2004-06-08 05:00:00  2004-06-08 03:00:00  MBSCL
7   2004-06-12 22:00:00  2004-06-13 12:00:00  2004-06-13 06:00:00    CCM
8   2004-06-13 03:00:00  2004-06-13 08:00:00  2004-06-13 06:00:00  MBCCM
9   2004-06-14 00:00:00  2004-06-14 03:00:00  2004-06-14 02:00:00  MBSCL
10  2004-06-14 03:00:00  2004-06-14 09:00:00  2004-06-14 07:00:00  MBSCL
11  2004-06-17 08:00:00  2004-06-17 14:00:00  2004-06-17 11:00:00  MBCCM
12  2004-06-17 12:00:00  2004-06-17 17:00:00  2004-06-17 14:00:00  MBCCM
13  2004-06-22 00:00:00  2004-06-22 08:00:00  2004-06-22 06:00:00    SCL
14  2004-06-22 08:00:00  2004-06-22 14:00:00  2004-06-22 11:00:00  MBCCM 
15  2004-06-22 23:00:00  2004-06-23 09:00:00  2004-06-23 06:00:00    CCM
16  2004-07-01 05:00:00  2004-07-01 09:00:00  2004-07-01 06:00:00  MBCCM
17  2004-07-02 00:00:00  2004-07-02 04:00:00  2004-07-02 02:00:00  MBSCL
18  2004-07-04 12:00:00  2004-07-04 15:00:00  2004-07-04 13:00:00  MBCCM
19  2004-07-06 04:00:00  2004-07-06 13:00:00  2004-07-06 07:00:00    SCL
20  2004-07-07 04:00:00  2004-07-07 12:00:00  2004-07-07 10:00:00    CCM
21  2004-07-08 03:00:00  2004-07-08 06:00:00  2004-07-08 05:00:00  MBCCM
22  2004-07-08 12:00:00  2004-07-08 17:00:00  2004-07-08 13:00:00  MBCCM
23  2004-07-08 02:00:00  2004-07-08 06:00:00  2004-07-08 04:00:00  MBCCM
24  2004-07-09 05:00:00  2004-07-09 12:00:00  2004-07-09 08:00:00    CCM
25  2004-07-11 18:00:00  2004-07-12 12:00:00  2004-07-11 21:00:00  MBSCL
26  2004-07-11 23:00:00  2004-07-12 05:00:00  2004-07-12 02:00:00  MBSCL
27  2004-07-15 11:00:00  2004-07-15 19:00:00  2004-07-15 12:00:00    CCM
28  2004-07-16 12:00:00  2004-07-16 16:00:00  2004-07-16 14:00:00  MBCCM
29  2004-07-17 02:00:00  2004-07-17 06:00:00  2004-07-17 05:00:00  MBCCM

现在我想获得一个小时内所有课程的出现情况。例如，在 Date_inicio、Date_fin 和 Date_max 中某个时间出现了多少次不同的类。从 df 我得到下一个频率表，

frec

       Frec_Inicio  Frec_Max  Frec_Fin
Horas                           
1           2             0    1
2           3             8    1
3           5             3    2
4           6             2    6
5           6             6    5
6           5             6    4
7           5             7    2
8           2             4    5
9           1             6    6 
10          0             3    2
11          2             5    5
12          4             1    9
13          2             4    2
14          3             2    4
15          0             2    3
16          1             1    3
17          0             2    3 
18          1             1    1
19          0             0    3
20          1             1    1
21          1             1    0
22          3             1    0
23          9             1    0
24          8             3    2

现在，我的目标是绘制如下图所示的 3D 条形

为此，我编写了以下代码

fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
xpos=np.arange(frec.shape[0])
ypos=np.arange(frec.shape[1])
xpos, ypos = np.meshgrid(xpos+0.25, ypos+0.25)
xpos = xpos.flatten()
ypos = ypos.flatten()
zpos=np.zeros(frec.shape).flatten()

dx=0.5 * np.ones_like(zpos)
dy=0.5 * np.ones_like(zpos)
dz=frec.values.ravel()
dz[np.isnan(dz)]=0.

ax.bar3d(xpos,ypos,zpos,dx,dy,dz,color='b', alpha=0.5)
ax.set_xticks([.5,1.5,2.5])
ax.set_yticks([.5,1.5,2.5,3.5])
ax.w_yaxis.set_ticklabels(frec.columns)
ax.w_xaxis.set_ticklabels(frec.index)
ax.set_xlabel('Time')
ax.set_ylabel('B')
ax.set_zlabel('Occurrence')
plt.show()

如何获得更好的情节，类似于上图？

【问题讨论】：

你能发布数据吗？我不明白你为什么要做crosstab 并从Forma 中删除所有列，也许你想要Series.value_counts。 matplotlib 不能做真正的 3D 绘图，试试visvis。
现在，我进行编辑以便更好地理解

标签： python matplotlib pandas

【解决方案1】：

这里是计数的代码：

import pandas as pd
text="""Date_inicio,             Date_Fin,             Date_Max,       Clase
2004-04-09 23:00:00,  2004-04-10 04:00:00,  2004-04-10 02:00:00,  MBCCM
2004-04-12 23:00:00,  2004-04-13 04:00:00,  2004-04-13 00:00:00,  MBSCL
2004-04-24 04:00:00,  2004-04-24 12:00:00,  2004-04-24 09:00:00,    SCL
2004-05-02 07:00:00,  2004-05-02 14:00:00,  2004-05-02 11:00:00,    SCL
2004-05-30 05:00:00,  2004-05-30 08:00:00,  2004-05-30 07:00:00,  MBCCM
2004-05-31 03:00:00,  2004-05-31 07:00:00,  2004-05-31 05:00:00,  MBCCM   
2004-06-08 00:00:00,  2004-06-08 05:00:00,  2004-06-08 03:00:00,  MBSCL
2004-06-12 22:00:00,  2004-06-13 12:00:00,  2004-06-13 06:00:00,    CCM
2004-06-13 03:00:00,  2004-06-13 08:00:00,  2004-06-13 06:00:00,  MBCCM
2004-06-14 00:00:00,  2004-06-14 03:00:00,  2004-06-14 02:00:00,  MBSCL
2004-06-14 03:00:00,  2004-06-14 09:00:00,  2004-06-14 07:00:00,  MBSCL
2004-06-17 08:00:00,  2004-06-17 14:00:00,  2004-06-17 11:00:00,  MBCCM
2004-06-17 12:00:00,  2004-06-17 17:00:00,  2004-06-17 14:00:00,  MBCCM
2004-06-22 00:00:00,  2004-06-22 08:00:00,  2004-06-22 06:00:00,    SCL
2004-06-22 08:00:00,  2004-06-22 14:00:00,  2004-06-22 11:00:00,  MBCCM 
2004-06-22 23:00:00,  2004-06-23 09:00:00,  2004-06-23 06:00:00,    CCM
2004-07-01 05:00:00,  2004-07-01 09:00:00,  2004-07-01 06:00:00,  MBCCM
2004-07-02 00:00:00,  2004-07-02 04:00:00,  2004-07-02 02:00:00,  MBSCL
2004-07-04 12:00:00,  2004-07-04 15:00:00,  2004-07-04 13:00:00,  MBCCM
2004-07-06 04:00:00,  2004-07-06 13:00:00,  2004-07-06 07:00:00,    SCL
2004-07-07 04:00:00,  2004-07-07 12:00:00,  2004-07-07 10:00:00,    CCM
2004-07-08 03:00:00,  2004-07-08 06:00:00,  2004-07-08 05:00:00,  MBCCM
2004-07-08 12:00:00,  2004-07-08 17:00:00,  2004-07-08 13:00:00,  MBCCM
2004-07-08 02:00:00,  2004-07-08 06:00:00,  2004-07-08 04:00:00,  MBCCM
2004-07-09 05:00:00,  2004-07-09 12:00:00,  2004-07-09 08:00:00,    CCM
2004-07-11 18:00:00,  2004-07-12 12:00:00,  2004-07-11 21:00:00,  MBSCL
2004-07-11 23:00:00,  2004-07-12 05:00:00,  2004-07-12 02:00:00,  MBSCL
2004-07-15 11:00:00,  2004-07-15 19:00:00,  2004-07-15 12:00:00,    CCM
2004-07-16 12:00:00,  2004-07-16 16:00:00,  2004-07-16 14:00:00,  MBCCM
2004-07-17 02:00:00,  2004-07-17 06:00:00,  2004-07-17 05:00:00,  MBCCM"""
import io
df = pd.read_csv(io.BytesIO(text), skipinitialspace=True)
df.drop(["Clase"], axis=1, inplace=True)
df = df.apply(lambda s:s.str[11:13]).convert_objects(convert_numeric=True)
df2 = df.apply(lambda s:s.value_counts())
print df2

这是绘制 3d 条的代码：

import pandas as pd
text="""Horas Frec_Inicio  Frec_Max  Frec_Fin
1           2             0    1
2           3             8    1
3           5             3    2
4           6             2    6
5           6             6    5
6           5             6    4
7           5             7    2
8           2             4    5
9           1             6    6 
10          0             3    2
11          2             5    5
12          4             1    9
13          2             4    2
14          3             2    4
15          0             2    3
16          1             1    3
17          0             2    3 
18          1             1    1
19          0             0    3
20          1             1    1
21          1             1    0
22          3             1    0
23          9             1    0
24          8             3    2"""
import io
df = pd.read_csv(io.BytesIO(text), skipinitialspace=True, delim_whitespace=True)
df.set_index("Horas", inplace=True)
columns_name = [x.replace("_", " ") for x in df.columns]
df.columns = [0, 2, 4]
x, y, z = df.stack().reset_index().values.T

import visvis as vv
app = vv.use()

f = vv.clf()
a = vv.cla()

bar =vv.bar3(x, y, z, width=0.8)
bar.colors = ["r","g","b"] * 24
a.axis.yTicks = dict(zip(df.columns, columns_name))
app.Run()

输出：

【讨论】：