如何在数据框应用Series.value_counts（）？

Question

在一个数据帧我想计数每一列的值，并使用该值作为指标。

我想关闭这个：

Q1                   Q2                  Q3
Strongly agree       Agree               Undecided
Undecided            Agree               More or less disagree
Strongly agree       Agree               Undecided
Strongly agree       Strongly Disagree   Disagree
More or less agree   Undecided           Strongly disagree

这个：

                        Q1  Q2  Q3
Strongly agree          3   0   0
Agree                   0   3   0
More or less agree      1   0   0
Undecided               1   1   2
More or less disagree   0   0   1
Disagree                0   0   1
Strongly disagree       0   1   1

这怎么可能有熊猫吗？

Answer 1

如果你坚持value_counts，您可以使用stack和groupby事先然后value_counts之前调用unstacking：

df.stack().groupby(level=[1]).value_counts().unstack(0, fill_value=0)

                       Q1  Q2  Q3
Agree                   0   3   0
Disagree                0   0   1
More or less agree      1   0   0
More or less disagree   0   0   1
Strongly Disagree       0   1   0
Strongly agree          3   0   0
Strongly disagree       0   0   1
Undecided               1   1   2

---

另一种选择是使用melt和pivot_table：

(df.melt()
   .pivot_table(columns='variable', index='value', aggfunc='size', fill_value=0))

variable               Q1  Q2  Q3
value                            
Agree                   0   3   0
Disagree                0   0   1
More or less agree      1   0   0
More or less disagree   0   0   1
Strongly Disagree       0   1   0
Strongly agree          3   0   0
Strongly disagree       0   0   1
Undecided               1   1   2

---

解决方案使用crosstab：

v = df.melt()
pd.crosstab(v['value'], v['variable'])

variable               Q1  Q2  Q3
value                            
Agree                   0   3   0
Disagree                0   0   1
More or less agree      1   0   0
More or less disagree   0   0   1
Strongly Disagree       0   1   0
Strongly agree          3   0   0
Strongly disagree       0   0   1
Undecided               1   1   2

Answer 2

您可以将pd.Series.value_counts功能对整个数据框，并填写NaN值0。

df.apply(pd.Series.value_counts,axis=0).fillna(0)