Question

我是Panda的新手，需要使用excel和数据透视表自动化之前提供的报告。我也不了解枢轴表。

因为我知道python我以为我可以使用它。我曾在csv-reader，csv-writer and openpyxl`工作过。

我认为panda可以用于数据分析，我可以用最少的代码完成这项工作。

而不是使用openpyxl或csv reader/writer一行一行但是，由于我是新手，我正在努力解决如何在panda

中完成这项工作

以下是数据框示例

编辑：文字格式

serial_number   item_name   status  date
1   foo1        done           2015-01-11
2   foo2        done           2016-01-11
3   foo3        not_done       2015-02-12
4   foo4        not_done       2016-01-12
5   foo5        on_hold        2015-03-13
6   foo6        not_done       2016-02-13
7   foo7        done           2016-03-14
8   foo7        done           2016-02-15
9   foo8        not_done       2016-03-16
10  foo8        on_hold        2016-04-17
11  foo9        on_hold        2016-04-18

从excel操作中，它提供以下报告

编辑：文字格式

        Status          
Year    Start_Month done    not_done    on_hold total
2015    January      1       0          0        1
        February     0       1          0        1
        March        0       0          1        1
2016    January      1       1          0        2
        February     1       1          0        2
        March        1       1          0        2
        April        0       0          2        2
Grand_Total          4       4          3        11

她试图为上述操作编写一个pivot_table

table = pd.pivot_table(df, values=["donel","not_done","on_hold"],\
 index=["date"], columns=["status"]

并且，这是我收到的错误消息

Traceback (most recent call last):
  File "<pyshell#5>", line 1, in <module>
table = pd.pivot_table(df, values=["Implementation - Successful","Closed Incomplete","Backed Out"], index=["chg_year","chg_month"], columns=["chg_state"]
  File "C:\Python27\lib\site-packages\pandas\tools\pivot.py", line 121, in pivot_table
    agged = grouped.agg(aggfunc)
  File "C:\Python27\lib\site-packages\pandas\core\groupby.py", line 3597, in aggregate
    return super(DataFrameGroupBy, self).aggregate(arg, *args, **kwargs)
  File "C:\Python27\lib\site-packages\pandas\core\groupby.py", line 3114, in aggregate
    result, how = self._aggregate(arg, _level=_level, *args, **kwargs)
  File "C:\Python27\lib\site-packages\pandas\core\base.py", line 428, in _aggregate
    return getattr(self, arg)(*args, **kwargs), None
  File "C:\Python27\lib\site-packages\pandas\core\groupby.py", line 964, in mean
    return self._cython_agg_general('mean')
  File "C:\Python27\lib\site-packages\pandas\core\groupby.py", line 3048, in _cython_agg_general
    how, numeric_only=numeric_only)
  File "C:\Python27\lib\site-packages\pandas\core\groupby.py", line 3094, in _cython_agg_blocks
    raise DataError('No numeric types to aggregate')
DataError: No numeric types to aggregate

Answer 1

下次，请以文本格式提供您的样本数据 - 或者更好的是，作为生成它的代码。

pd.crosstab

使用pd.pivot_table（而不是output = pd.crosstab(df['date'], df['status']).resample('M').sum().dropna()）并按月重新采样。

output['total'] = output.sum(axis=1)

计算每行的总数。

dates = output.index.to_series()
output.index = pd.MultiIndex.from_arrays(
    [dates.dt.year, dates.dt.strftime('%B')],
    names=['Year', 'Start_Month'])
print(output)

# status            done  not_done  on_hold  total
# Year Start_Month                                
# 2015 January       1.0       0.0      0.0    1.0
#      February      0.0       1.0      0.0    1.0
#      March         0.0       0.0      1.0    1.0
# 2016 January       1.0       1.0      0.0    2.0
#      February      1.0       1.0      0.0    2.0
#      March         1.0       1.0      0.0    2.0
#      April         0.0       0.0      2.0    2.0

重新索引以获得格式化的年份和月份，如您所需的输出。

grand_total = output.sum()
print(grand_total)

# status
# done         4.0
# not_done     4.0
# on_hold      3.0
# total       11.0
# dtype: float64

按列列出的总数在同一数据框中没有位置。

     F = 2 * 3 * a * b * 7&
    & * 3 * b * c&
     AA = ...

Answer 2

试试这个：

    #break out month and year
    df.loc[:, 'month'] = df['date'].apply(lambda x: x.month)
    df.loc[:, 'year'] = df['date'].apply(lambda x: x.year)
    #aggregate with column headers as 'status' values
    df.pivot_table(index = ['month', 'year'], columns = ['status'], values = 'item_name', aggfunc = 'count')

的产率：

status      done  not_done
month year                
1     2016   1.0       NaN
2     2016   NaN       1.0
3     2016   1.0       NaN
4     2016   NaN       1.0
5     2016   1.0       NaN
6     2016   NaN       1.0
7     2016   1.0       NaN
8     2016   1.0       NaN
9     2016   1.0       NaN
10    2016   1.0       NaN

熊猫：每个数据透视表的数据处理

2 个答案: