Pandas multiconditional drop using two dataframes

Question

How can I drop some values from a dataframe using another dataframe as parameter?

df1
code | reply
A1   | yes
A2   | yes
A3   | no

df2
code |
A1   |
A1   |
A3   |

df_new = df1.drop(df1['code'] == df2['code'] & df1['reply'] != 'yes')

df_new
code | reply
A1   | yes

Is there a simple way to do this using .drop()?

There are several ways, have a look at those answers: https://stackoverflow.com/questions/33282119/pandas-filter-dataframe-by-another-dataframe-by-row-elements — tomasborrella, Jan 10 '23 at 14:25

score 2 · Accepted Answer · answered Jan 10 '23 at 14:27

2

out = df1[df1['code'].isin(df2['code']) & df1['reply'].eq('yes')]

Output:

  code reply
0   A1   yes

answered Jan 10 '23 at 14:27

mozway

Is there a "inverse" of ```.eq()``` in pandas? I'll use it too – null92 Jan 10 '23 at 14:35
1

`ne()` is the opposite of `eq()` – Corralien Jan 10 '23 at 14:36
@null92. For your information `AND` and `&` are not the same thing. – Corralien Jan 10 '23 at 14:37
@mozway. In fact, I will removed my answer. It does not make sense to use `merge` here – Corralien Jan 10 '23 at 14:41

Corralien · Answer 2 · 2023-01-10T14:40:26.360

1

The logic is unclear but you can accomplish what you want without the use of drop:

>>> df1[df1['reply'] == 'yes'].merge(df2.drop_duplicates('code'))

  code reply
0   A1   yes

edited Jan 10 '23 at 14:40

answered Jan 10 '23 at 14:27

Corralien

actually your drop and my indexing are equivalent using [De Morgan's law](https://en.wikipedia.org/wiki/De_Morgan%27s_laws), so yes I always try to keep the most simple variant in such cases ;) – mozway Jan 10 '23 at 14:34
1

That's why I'm pointing it out to the OP. – Corralien Jan 10 '23 at 14:36
I was about to add this variant and the wikipedia link in my answer so I commented instead as you had just edited ;) – mozway Jan 10 '23 at 14:38
OK. I removed this part from my answer. – Corralien Jan 10 '23 at 14:39

2 Answers2