输出groupby到csv文件熊猫

Question

I have a sample dataset:

import pandas as pd
df = {'ID': ['H1','H2','H3','H4','H5','H6'],
      'AA1': ['C','B','B','X','G','G'],
      'AA2': ['W','K','K','A','B','B'],
      'name':['n1','n2','n3','n4','n5','n6']
}

df = pd.DataFrame(df)

it looks like :

df
Out[32]: 
   AA1 AA2  ID name
0   C   W  H1   n1
1   B   K  H2   n2
2   B   K  H3   n3
3   X   A  H4   n4
4   G   B  H5   n5
5   G   B  H6   n6

I want to groupby AA1 and AA2 (unique AA1 and AA2 pair) and it doesn't matter which ID and name values the unique pair picks along with it, and output that to a .csv file, so the output in the .csv file would look like:

 AA1 AA2  ID name
  C   W  H1   n1
  B   K  H2   n2
  X   A  H4   n4
  G   B  H5   n5

i tried the code:

df.groupby('AA1','AA2').apply(to_csv('merged.txt', sep = '\t', index=False))

but the to_csv was not recognized, what can i put in the .apply() to just output the groupby results to a csv file?

Answer 1

The problem is that you are trying to apply a function to_csv which doesn't exist. Anyway, groupby also doesn't have a to_csv method. pd.Series and pd.DataFrame do.

What you should really use is drop_duplicates here and then export the resulting dataframe to csv:

df.drop_duplicates(['AA1','AA2']).to_csv('merged.txt')

---

PS: If you really wanted a groupby solution, there's this one that happens to be 12 times slower than drop_duplicates...:

df.groupby(['AA1','AA2']).agg(lambda x:x.value_counts().index[0]).to_csv('merged.txt')

Answer 2

you can use groupby with head

df.groupby(['AA1', 'AA2']).head(1)