无组织数据帧创建的条形图？

Question

从下表中，我需要创建4个不同的条形图，对应于4个不同的位置TST1{}{}{}{}

如果可能的话，每个条形图应该有8个记号，用于NOT_DONE{}{}{}{}{}{}{}按该顺序排列。刻度将对应于给定位置的每个“值”出现的时间数。（位置是4个选项之一：L1{}{}{}）

但是：

仅考虑最右边列中的值，这意味着如果在TST5中未找到值，则程序应检查TST4等，直到找到值为止。如果在这5列中未找到任何值，则不计算任何值。如果找到一个值，那么它与它左边的内容无关

我的思想过程是创建一个新的列数据框架，其中包含我需要的值（因此每行的值都是最正确的）及其对应的位置。我对这一切都是新手，不知道如何做到这一点，所以我非常感谢你对我的帮助

我需要使用Python2.7，我还使用seaborn进行绘图

+-------+----------+----------+----------+--------+----------+
| PLACE | TST1     | TST2     | TST3     | TST4   | TST5     |
+-------+----------+----------+----------+--------+----------+
| L1    |          | NOT_DONE |          |        | 50       |
+-------+----------+----------+----------+--------+----------+
| L1    |          |          | 35       |        |          |
+-------+----------+----------+----------+--------+----------+
| L4    |          |          |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L3    |          |          | INCOMP   |        |          |
+-------+----------+----------+----------+--------+----------+
| L2    | UNTESTED |          |          | INCOMP |          |
+-------+----------+----------+----------+--------+----------+
| L3    |          |          |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L4    |          | 30       |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L3    |          | INCOMP   | 40       |        |          |
+-------+----------+----------+----------+--------+----------+
| L4    |          |          |          |        | UNTESTED |
+-------+----------+----------+----------+--------+----------+
| L1    |          |          |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L3    |          | INCOMP   |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L2    |          |          |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L2    |          | 50       |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L3    |          |          | UNTESTED | 35     | NOT_DONE |
+-------+----------+----------+----------+--------+----------+
| L1    |          |          |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L2    |          | 40       |          | INCOMP |          |
+-------+----------+----------+----------+--------+----------+
| L3    |          |          |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L1    |          |          |          |        |          |
+-------+----------+----------+----------+--------+----------+
| L4    |          | NOT_DONE |          | 30     | NOT_DONE |
+-------+----------+----------+----------+--------+----------+

Answer 1

I am required to use python 2.7, I am also using seaborn for the plotting.

在Python2.7.18和pandas 0.24.2上测试（尽管在Python3中工作良好）：

1. 使用^{}沿columns传播最右边的值（忽略PLACE）：

   df['TST'] = df.drop(columns='PLACE').ffill(axis='columns').iloc[:, -1]
   

2. 按PLACE分组并获取其^{}：

   data = df.groupby('PLACE')['TST'].value_counts().reset_index(name='COUNT')
   
   #   PLACE       TST  COUNT
   # 0    L1        35      1
   # 1    L1        50      1
   # 2    L2    INCOMP      2
   # 3    L2        50      1
   # 4    L3    INCOMP      2
   # 5    L3        40      1
   # 6    L3  NOT_DONE      1
   # 7    L4        30      1
   # 8    L4  NOT_DONE      1
   # 9    L4  UNTESTED      1
   

3. 然后将此data传递到^{}（使用order参数设置首选的勾号顺序）：

   incompletes = ['NOT_DONE', 'INCOMP', 'UNTESTED']
   ticks = incompletes + sorted(data.TST.unique())[:len(incompletes)]
   
   g = sns.catplot(x='TST', y='COUNT', col='PLACE', col_wrap=2,
                   data=data, order=ticks, kind='bar')
   g.set_xticklabels(rotation=90)
   

   

---

版本：

>>> sys.version
2.7.18 (default, Mar 15 2021, 14:29:03) \n[GCC 10.2.0]
>>> pandas.__version__
0.24.2
>>> matplotlib.__version__
2.2.5
>>> seaborn.__version__
0.9.1