CODEWITHSUNDEEP
pythonmergefuzzy-searchdifflib
user2649353
user2649353

Reputation: 377

Using difflib SequenceMatcher ratio to merge in Pandas

I'm trying to figure out if there's a way to do fuzzy merges of string in Pandas based on the difflib SequenceMatcher ration. Basically, I have two dataframes that look like this:

df_a
company    address        merged
Apple     PO Box 3435       1

df_b
company     address
Apple Inc   PO Box 343

And I want to merge like this:

df_c = pd.merge(df_a, df_b, how = 'left', on = (difflib.SequenceMatcher(None, df_a['company'], df_b['company']).ratio() > .6) and (difflib.SequenceMatcher(None, df_a['address'], df_b['address']).ratio() > .6)

There are a few posts that are close to what I'm looking for, but none of them work with what I want to do. Any suggestions on how to do this kind of fuzzy merge using difflib?

Upvotes: 1

Views: 1963

Answers (1)

dermen
dermen

Reputation: 5382

Something that might work: test for partial matches for all combinations of column values. If there is a match assign a key to df_b for merging

df_a['merge_comp'] = df_a['company'] # we will use these as the merge keys
df_a['merge_addr'] = df_a['address']

for comp_a, addr_a in df_a[['company','address']].values:
    for ixb, (comp_b, addr_b) in enumerate(df_b[['company','address']].values)
        if difflib.SequenceMatcher(None,comp_a,comp_b).ratio() > .6:
            df_b.ix[ixb,'merge_comp'] = comp_a # creates a merge key in df_b
        if difflib.SequenceMatcher(None,addr_a, addr_b).ratio() > .6:
            df_b.ix[ixb,'merge_addr'] = addr_a # creates a merge key in df_b

Now you can merge

merged_df = pandas.merge(df_a,df_b,on=['merge_addr','merge_comp'],how='inner')

Upvotes: 2

Related Questions