CODEWITHSUNDEEP
pythonpandas
dward4
dward4

Reputation: 1922

Replace values in dataframe column if second column matches a given list pandas

I have the following dataframe

      col1          col2
0     str9          val1
1     str8          val2
2     str4          val3
3     str2          val4
4  unknown1    asdvstr1df
5  random1  teststr2test

and the following list

strings = ['str1', 'str2', 'str3', 'str4']

I want to replace the value in column 1 if there is a match at any point in the string in column two with the given list, strings.

Note: col2 can have the values of the list 'string' inside the string, or at either end as well.

Currently I am doing this with an ugly and slow nested loop: 

import pandas as pd

data_file = pd.DataFrame(data = ([['str9', 'val1'], ['str8',  'val2'], ['str4','val3'] , ['str2', 'val4'] , ['unknown', 'asdvstr1df'] , ['unknown', 'teststr2test']] ), columns = (['col1', 'col2']), dtype = str)
strings = ['str1', 'str2', 'str3', 'str4']

for value in range(data_file.shape[0]):
    for text in strings:
        if (str(data_file.col2[value]).find(text) != -1):
            data_file.loc[value, 'col1'] = text

I am unsure how to improve upon this slow process. How can I make this run faster than the current O(nm) time (n is size of the data_file, m the size of the list called strings)?

The output should be:

   col1          col2
0  str9          val1
1  str8          val2
2  str4          val3
3  str2          val4
4  str1    asdvstr1df
5  str2  teststr2test

Upvotes: 2

Views: 130

Answers (3)

Scott Boston
Scott Boston

Reputation: 153460

IIUC,

x = '(' + '|'.join(strings)+ ')'
df.assign(col1 = df.col2.str.extract(x, expand=False).combine_first(df.col1))

Output:

   col1          col2
0  str9          val1
1  str8          val2
2  str4          val3
3  str2          val4
4  str1    asdvstr1df
5  str2  teststr2test

Upvotes: 5

BENY
BENY

Reputation: 323226

You can using replace twice with regex

d1=dict(zip(strings,[1,2,3,4]))
d2=dict(zip([1,2,3,4],strings))

df.loc[df.col1=='unknown','col1']=df.col2.replace(d1,regex=True).replace(d2)
df
Out[970]: 
   col1          col2
0  str9          val1
1  str8          val2
2  str4          val3
3  str2          val4
4  str1    asdvstr1df
5  str2  teststr2test

Upvotes: 3

FadeoN
FadeoN

Reputation: 101

Try this: 

data_file["col1"] = data_file["col2"].apply(lambda y:strings[[True if x in y 
else False for x in strings ].index(True)] if any([True if x in y else False 
for x in strings ]) else y)
print(data_file)

OUTPUT:

   col1          col2
0  val1          val1
1  val2          val2
2  val3          val3
3  val4          val4
4  str1    asdvstr1df
5  str2  teststr2test

Upvotes: 1

Related Questions