12

I have a pandas dataframe as follows:

A   B   C
1   2   x
1   2   y
3   4   z
3   5   x

I want that only 1 row remains of rows that share the same values in specific columns. In the example above I mean columns A and B. In other words, if the values of columns A and B occur more than once in the dataframe, only one row should remain (which one does not matter).

FWIW: the maximum number of so called duplicate rows (that is, where column A and B are the same) is 2.

The result should looke like this:

A   B   C
1   2   x
3   4   z
3   5   x

or

A   B   C
1   2   y
3   4   z
3   5   x
beta
  • 5,324
  • 15
  • 57
  • 99

1 Answers1

26

Use drop_duplicates with parameter subset, for keeping only last duplicated rows add keep='last':

df1 = df.drop_duplicates(subset=['A','B'])
#same as
#df1 = df.drop_duplicates(subset=['A','B'], keep='first')
print (df1)
   A  B  C
0  1  2  x
2  3  4  z
3  3  5  x

df2 = df.drop_duplicates(subset=['A','B'], keep='last')
print (df2)
   A  B  C
1  1  2  y
2  3  4  z
3  3  5  x
jezrael
  • 822,522
  • 95
  • 1,334
  • 1,252
  • @ jezrael If I want to remove duplicates just from a column without removing rows. Let's say I have 10 rows which is corrowponds to same time instant . So I want to write to txt file but only one time instant I want to print for all 10 rows instead of showing same time for each rows. – Poka Aug 12 '18 at 06:51
  • 1
    @Poka - If dont want remove rows, only solution is replace duplicated values to `NaN` or `empty string`. Something like [`this solution`](https://stackoverflow.com/a/47553940/2901002) – jezrael Aug 12 '18 at 06:56