CODEWITHSUNDEEP
pythonregexpython-2.7python-3.xpandas
Shubham R
Shubham R

Reputation: 7644

Split Column containing 2 values into different column in pandas df

i have a table in pandas df

bigram         frequency
(123,3245)       2
(676,35346)      84
(93,32)          9

and so on, till 50 rows.

what i am looking for is, split the bigram column into two different columns removing the brackets and comma like,

col1     col2      frequency
123       3245        2
676       35346       84
93        32          9

is there any way to split if after comma,and removing brackets.

Upvotes: 3

Views: 1488

Answers (3)

piRSquared
piRSquared

Reputation: 294488

very close to @Psidom's answer.
I use pd.DataFrame(df.bigram.values.tolist(), columns=['c1', 'c2']) instead of df.bigram.apply(lambda x: pd.Series(x, index=['col1', 'col2']))

pd.concat([pd.DataFrame(df.bigram.values.tolist(), columns=['c1', 'c2']),
           df.drop('bigram', 1)],
          axis=1)

enter image description here

Upvotes: 2

James
James

Reputation: 36691

Try creating a new column for each pair of the tuple.

df['col1'] = df['bigram'].apply(lambda x: x[0])
df['col2'] = df['bigram'].apply(lambda x: x[1])

To create a data frame with ONLY col1, col2, and frequency where the order is important, it is easier to create a new data frame altogether and populate it.

df_new = pd.DataFrame()
df_new['col1'] = df['bigram'].apply(lambda x: x[0])
df_new['col2'] = df['bigram'].apply(lambda x: x[1])
df_new['frequency'] = df['frequency']

Upvotes: 2

akuiper
akuiper

Reputation: 215057

If your bigram column happens to be string format, you can use .str.extract() method with regex to extract numbers from it:

pd.concat([df.bigram.str.extract('(?P<col1>\d+),(?P<col2>\d+)'), df.frequency], axis = 1)

enter image description here

Or if the bigram column is of tuple type:

Method1: use pd.Series to create columns from the tuple:

pd.concat([df.bigram.apply(lambda x: pd.Series(x, index=['col1', 'col2'])), 
           df.frequency], axis=1)

Method2: use .str to get the first and second element from the tuple

df['col1'], df['col2'] = df.bigram.str[0], df.bigram.str[1]
df = df.drop('bigram', axis=1)

Upvotes: 3

Related Questions