[英]Pandas Cumulative Sum in GroupBy
I have a time series data with symbol and their respective values at a particular time. 我有一个带符号的时间序列数据及其在特定时间的各自值。
index,symbol,value
01:00,A,10
01:00,B,15
01:01,A,15
01:01,B,25
01:02,A,30
01:02,B,45
Now I want create a 4th column, which has cumulative value on time series basis for each symbol but from each cumulative row, the first row value would be subtracted for each symbol respectively 现在我想创建第4列,其中每个符号的时间序列累计值,但从每个累积行,将分别减去每个符号的第一行值
index,symbol,value,adjustedCumulativeSum
01:00,A,10,0
01:00,B,15,0
01:01,A,15,15
01:01,B,25,25
01:02,A,30,45
01:02,B,45,70
I know how to do normal cumulative sum 我知道如何做正常的累计和
df = df.reset_index().sort_values(['index','symbol'])
df['cumlativesum'] = df.groupby('symbol')['value'].cumsum()
df = df.set_index('index')
But do I deduct row 0 value from all cumulative sums? 但是,我是否从所有累计金额中扣除第0行的值?
Use groupby
with custom function with cumsum
and substract first value selected by iat
: 将
groupby
与带有cumsum
自定义函数一起cumsum
并减去iat
选择的第一个值:
df['adjustedCumulativeSum']=df.groupby('symbol')['value'].apply(lambda x:x.cumsum()-x.iat[0])
print (df)
index symbol value adjustedCumulativeSum
0 01:00 A 10 0
1 01:00 B 15 0
2 01:01 A 15 15
3 01:01 B 25 25
4 01:02 A 30 45
5 01:02 B 45 70
You can subtract the first value (extracted with .iat[0]
) for each group in a transform
function: 您可以在
transform
函数中为每个组减去第一个值(使用.iat[0]
提取):
df['cumlativesum'] = df.groupby('symbol')['value'].transform(lambda g: g.cumsum()-g.iat[0])
df = df.set_index('index')
df
# symbol value cumlativesum
#index
#01:00 A 10 0
#01:00 B 15 0
#01:01 A 15 15
#01:01 B 25 25
#01:02 A 30 45
#01:02 B 45 70
df.groupby('sy').val.apply(lambda x : x.cumsum()-x.values.tolist()[0])
Out[907]:
0 0
1 0
2 15
3 25
4 45
5 70
Name: val, dtype: int64
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.