熊猫根据列总和的百分比分隔数据框

Question

Suppose I have a dataframe sorted from smallest to largest over ratio column, as below: (actual dataframe has thousands of rows)

identifier total ratio
1          15     0.21
2          500    0.21
3          70     0.56
4          200    0.75
5          540    0.99

and a cutoff value of:

cutoff = .3 

and at the end I want two csv files, one with the 30% of the sum of total with the lowest ratio (type1.csv), and one with the remaining 70% (type2.csv)

So far I have tried just taking the first 30% of the rows, as well as trying to multiply the

total * ratio 

and sorting on that new column, neither resulted in the correct lists at the end...

How to I assign weights for the total column's value, but then cut on the ratio column?

Answer 1

Like this?

cols = ['identifier', 'total', 'ratio']

data = [
[1          ,15    , 0.21],
[2          ,500    ,0.21],
[3          ,70     ,0.56],
[4          ,200    ,0.75],
[5          ,540    ,0.99]
]
import pandas as pd
df = pd.DataFrame(data=data, columns=cols)

df['s']=(df.total*df.ratio).cumsum()
df['cutoff']=df.s/df.s.iloc[-1]

type1 = df[df['cutoff'] < 0.3]
type1[['identifier', 'total', 'ratio']].to_csv(index=False, path_or_buf='type1.csv')


type2 = df[df['cutoff'] >= 0.3]
type2[['identifier', 'total', 'ratio']].to_csv(index=False, path_or_buf='type2.csv')