CODEWITHSUNDEEP
pythonpandas
Andy Do
Andy Do

Reputation: 99

Return value from other dataframe from partial string match

I'm trying create a new dataframe column that is a partial string match from another dataframe. How would I do the example below? 

df1:
#   id
1   666666
2   666667
3   666668
4   666667

df2
#   ref
1   ref_666666_blah blah
2   ref_666667_blah blah
3   ref_666668_blah blah
4   ref_666667_blah blah

df3 #what I want
#   id      match
1   666666  ref_666666_blah blah
2   666667  ref_666667_blah blah
3   666668  ref_666668_blah blah
4   666667  ref_666667_blah blah

I know this is not the code but I'm trying to do the below:

df1['match'] = df2['ref'].map(lambda x: x if x.str.contains(df1['match'])

Thanks!

Upvotes: 0

Views: 102

Answers (1)

hilberts_drinking_problem
hilberts_drinking_problem

Reputation: 11602

There are a number of ways to accomplish this.

If you are able to extract the id from the ref column, as you would in this particular example by df2[id] = df2.ref.apply(lambda c: c.split('_')[1]), you could proceed with df1.join(df2, on = 'id').

If you need to call some more complicated match function, you could do the following:

def getMatch(str_id):
    matches = (c for c in df2['ref'] if str_id in c)
    try:
        return matches.next()
    except:
        return None

df1['match'] = df1['id'].apply(getMatch)

This will result in a number of redundant comparisons, so you should think if there are relationships in your data that can simplify the match. For instance, if each ref matches to at most one id or if you can somehow sort both DataFrames in a meaningful way and merge them recursively.

Upvotes: 1

Related Questions