SettingWithCopyWarning, while mapping with a dictionary

Question

I'm currently try to understand this warning:

SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

I've seen this alot on SO, however my issue arises when trying to map.

My code is as such:

def merger(df):

     qdf = pd.read_csv('domains_only_df.csv')

     unfilt_rel_domains = qdf[['name', 'hits', 'owner', 'curated', 'domain']].copy()
     rel_domains = unfilt_rel_domains[unfilt_rel_domains['curated'] == 0]

     hits_dict= pd.Series(rel_domains.hits.values, index=rel_domains.owner).to_dict()

     name_dict = pd.Series(rel_domains.name.values, index=rel_domains.owner).to_dict()

     domain_dict = pd.Series(rel_domains.domain.values, index=rel_domains.owner).to_dict()

     df['Hits'] = df['eid'].map(hits_dict).fillna(0)
     df['Existing_domain'] = df['eid'].map(name_dict).fillna(0)
     df['idn'] = df['eid'].map(domain_dict).fillna(0)

return df

The error occurs with .map(), my question is how would write a mapping using the warning's recommendation of using .loc[row_indexer,col_indexer] = value? I need .map() for speed and the lookup but I'm not quite sure how to avoid this warning.

You've shown `unfilt_rel_domains` and `rel_domains` but what is `df`? — cs95, Jun 18 '19 at 15:30
Sorry about that I'll adjust its part of a function, `df` is my main dataframe that is passed to to the function and the dicts are mapped to the original `df` — Sebastian Goslin, Jun 18 '19 at 15:33
the warning is because you call the function `merge` on a slice of some bigger dataframe and try to modify the values of `df`, not because you use the `map` function. — Quang Hoang, Jun 18 '19 at 15:53
@QuangHoang so would I just need to set something like `df.copy()`? — Sebastian Goslin, Jun 18 '19 at 15:56
since you are returning `df`, yes, you can do `df = df.copy()` at the beginning of the function. — Quang Hoang, Jun 18 '19 at 15:57

score 0 · Answer 1 · answered Jun 18 '19 at 15:59

0

Literally a one liner:

prior to doing anything: set df = df.copy() as to get an actually view as opposed to woking on another copy.

answered Jun 18 '19 at 15:59

Sebastian Goslin

477
1
3
22

SettingWithCopyWarning, while mapping with a dictionary

1 Answers1