Question

我有一个像下面这样的数据框

Index     Toronto_2000   Toronto_2001    Toronto_2002    Toronto_2003    Montreal_2000  Montreal_2001   Montreal_2002   Montreal_2003
ID:1012          100            98              102             105             101             104             108            110

我该如何计算每年的变化百分比以及每个城市的变化百分比？

由于存在不同的城市，因此像2001、2002、2003一样计算df.pct_change（）

Answer 1

我建议先用str.split将unstack重塑DataFrame，然后将groupby与pct_change一起使用：

df.columns = df.columns.str.split('_', expand=True)
df = df.unstack().reset_index()
df.columns = ['city','year','index','val']
print (df)
       city  year    index  val
0   Toronto  2000  ID:1012  100
1   Toronto  2001  ID:1012   98
2   Toronto  2002  ID:1012  102
3   Toronto  2003  ID:1012  105
4  Montreal  2000  ID:1012  101
5  Montreal  2001  ID:1012  104
6  Montreal  2002  ID:1012  108
7  Montreal  2003  ID:1012  110

df['pct'] = df.groupby('city')['val'].apply(lambda x: x.pct_change()) 
print (df)
       city  year    index  val       pct
0   Toronto  2000  ID:1012  100       NaN
1   Toronto  2001  ID:1012   98 -0.020000
2   Toronto  2002  ID:1012  102  0.040816
3   Toronto  2003  ID:1012  105  0.029412
4  Montreal  2000  ID:1012  101       NaN
5  Montreal  2001  ID:1012  104  0.029703
6  Montreal  2002  ID:1012  108  0.038462
7  Montreal  2003  ID:1012  110  0.018519

Answer 2

这是很长的路要走，也许专业人士会为您提供漂亮的衬里纸

df2 = df.T.reset_index()
df2['city'] = [x[0] for x in df2['index'].str.split('_')]
df2['year'] = [x[1] for x in df2['index'].str.split('_')]
df2['pct'] = df2.sort_values('year').groupby(['city'])['ID:1012'].pct_change()

输出

Index          index  ID:1012      city  year       pct
0       Toronto_2000      100   Toronto  2000       NaN
1       Toronto_2001       98   Toronto  2001 -0.029703
2       Toronto_2002      102   Toronto  2002 -0.019231
3       Toronto_2003      105   Toronto  2003 -0.027778
4      Montreal_2000      101  Montreal  2000  0.010000
5      Montreal_2001      104  Montreal  2001  0.061224
6      Montreal_2002      108  Montreal  2002  0.058824
7      Montreal_2003      110  Montreal  2003  0.047619

如何根据不同的年份和column_name

2 个答案:

输出