计算pandas数据帧行之间的百分比差异

Question

region  year      val
1.0     2015.0    6.775457e+05
1.0     2016.0    6.819761e+05
1.0     2017.0    6.864065e+05
2.0     2015.0    6.175457e+05
2.0     2016.0    6.419761e+05
3.0     2017.0    6.564065e+05

In the dataframe above, I want to compute the percentage difference between consecutive rows but only for the same region values. I tried this but not sure if it works. What is best way to achieve it?

df.groupby(['region', 'year'])['val'].pct_change()

Answer 1

You can use DataFrameGroupBy.pct_change with groupby by column region :

df['new'] = df.groupby('region')['val'].pct_change()
print (df)
   region    year       val       new
0     1.0  2015.0  677545.7       NaN
1     1.0  2016.0  681976.1  0.006539
2     1.0  2017.0  686406.5  0.006496
3     2.0  2015.0  617545.7       NaN
4     2.0  2016.0  641976.1  0.039560
5     3.0  2017.0  656406.5       NaN