Pandas dataframe groupby删除列

Question

I stumble upon with a problem with dataframe. I am using this snippet code to generate dataframe after that I group by dataframe based on 'chr' Column .

import pandas as pd

DF = pd.DataFrame({'chr':["chr3","chr3","chr7","chr6","chr1", "chr7"],'y':[10,20,30,40,50,90],'ds': 
     ['2018-01-01', '2018-01-02', '2018-01-01', '2018-01-01', '2018-01-01', '2018-12-01']})

DF.head(n=10)

    chr     y       ds
0   chr3    10  2018-01-01
1   chr3    20  2018-01-02
2   chr7    30  2018-01-01
3   chr6    40  2018-01-01
4   chr1    50  2018-01-01
5   chr7    90  2018-12-01


ans = [pd.DataFrame(y) for x, y in DF.groupby('chr', as_index=False)]
ans

[    chr   y          ds
4  chr1  50  2018-01-01,
     chr   y          ds
0  chr3  10  2018-01-01
1  chr3  20  2018-01-02,
     chr   y          ds
3  chr6  40  2018-01-01,
     chr   y          ds
2  chr7  30  2018-01-01
5  chr7  90  2018-12-01]

Please note that once I use groupby I store the result in list. As a result, I have list with nested dataframe based on chr . What is the way if I need to delete chr column in each sub dataframe from my list? I need simply to drop chr in each dataframe from the list. Please note that solution should scale on bigger list size.

Answer 1

You can do it while creating your original list like this if there are only two columns:

ans = [pd.DataFrame(y, columns=DF.columns.difference(['chr'])) for x, y in DF.groupby('chr', as_index=False)]    

Alternatively, drop chr from each subDf explicitly:

ans = [pd.DataFrame(y).drop('chr', axis=1) for x, y in DF.groupby('chr', as_index=False)]    

---

If you can't drop while creating the original list (as shown above), you can update it like this:

# Create `ans` as you're currently doing:
ans = [pd.DataFrame(y) for x, y in DF.groupby('chr', as_index=False)] 
#
# some processing on `ans`
#
# Now update `ans` by dropping "chr" from each subDf
ans = [df.drop('chr', axis=1) for df in ans]

Answer 2

This will drop the column chr during saving:

ans = [pd.DataFrame(y).drop('chr', axis=1) for x, y in DF.groupby('chr', as_index=False)]