问题
I have a dataframe with a Week , Campaign , Placement and Count column.
In order to compare counts per weeks by Campaign and Placement I created a pivot table that works great. How do I create a new column with the difference between these 2 weeks (in percentage if possible)?
Code:
dfPivot = pd.pivot_table(dfPivot, values='Count',\
index=['Campaign', 'Placement'],columns=['Week'], aggfunc=np.sum)
Current Output:
Week 2019-10-27 2019-11-03
Campaign Placement Code
A 111111111 4288.0 615.0
111111112 243.0 11.0
111111113 598.0 30.0
111111114 1041.0 377.0
111111115 7759.0 161.0
B 111111111 1252.0 241.0
111111112 643.0 124.0
111111113 135.0 30.0
111111114 8753.0 2327.0
111111115 7242.0 112.0
Expected Output:
Week 2019-10-27 2019-11-03 Difference
Campaign Placement Code
A 111111111 4288.0 615.0 -85.7%
111111112 243.0 11.0 -95.4%
111111113 598.0 30.0 -94.9%
111111114 1041.0 377.0 [...]
111111115 7759.0 161.0 [...]
B 111111111 1252.0 241.0 [...]
111111112 643.0 124.0 [...]
111111113 135.0 30.0 [...]
111111114 8753.0 2327.0 [...]
111111115 7242.0 112.0 [...]
Thank you!
回答1:
Use DataFrame.pct_change with selecting last row by positions and multiple by 100 for percentages:
df['diff'] = df.pct_change(axis=1).iloc[:, -1].mul(100)
print (df)
2019-10-27 2019-11-03 diff
Campaign Placement Code
A 111111111 4288.0 615.0 -85.657649
111111112 243.0 11.0 -95.473251
111111113 598.0 30.0 -94.983278
111111114 1041.0 377.0 -63.784822
111111115 7759.0 161.0 -97.924990
B 111111111 1252.0 241.0 -80.750799
111111112 643.0 124.0 -80.715397
111111113 135.0 30.0 -77.777778
111111114 8753.0 2327.0 -73.414829
111111115 7242.0 112.0 -98.453466
来源：https://stackoverflow.com/questions/59159650/take-difference-between-pivot-table-columns-in-python