I have a dataframe df with a column A.
There is a given string s.
I want to subset the dataframe with rows which have column values in A which are substrings of
given string s.
If I wanted the other way around, I would have done something like df[df['A'].contains(s)].
But I want to filter out rows where s is a superstring of value in df['A'].
I have not been able to find an answer to this hence apologies in advance if a duplicate question exists.
I've created a sample example:
s = 'sas'
df = pd.DataFrame(data={'A':['s', 'a', 'sa', 'da']})
The solution is:
filtered_df = df[[s.find(i)!=-1 for i in df['A'].values]]
It isn't a pandas way, but it works. I hope it will be helpful.
Output:
A
0 s
1 a
2 sa
UPD: If you don't want to add rows where s==i, you can modify this code like:
df[[((s.find(i)!=-1) and (s!=i)) for i in df['A'].values]]
s.find(i)!=-1 with i in s also works.I don't think you can do better than a list comprehension:
df[[x in s for x in df.A]]
However, if df.A is repetitive (e.g. a categorical) then you could optimize it
key, unique_vals = pd.factorize(df.A)
mask = np.asarray([x in s for x in unique_vals])
df.loc[mask[key]]
Or if df.A is a categorical
key, unique_vals = df.A.cat.codes, df.A.cat.categories
mask = np.asarray([x in s for x in unique_vals])
df.loc[mask[key]]
