Pandas Dataframe Column Value Conversion from string to binary

Question

I am trying to convert string values in a column to binary values.For example in the below table - the mapping will be like this {'cinema':0, 'education':1}.

I have defined a small function as below -

def numconv(a):
return a.map({'education' : 1,'cinema' : 0})

and then I have used 'apply' function to update the values in the dataframe -

train_docs['Class'] = train_docs['Class'].apply(numconv)

But I am getting error as below -

What I am doing wrng here?It should work as per my understanding.

This would likely be faster to do in a vectorized fashion if its a large dataframe e.g. `df['Class]=np.where((df['Class']=='education'),1,0)` — brobertsp, Apr 23 '21 at 11:05

score 1 · Answer 1 · answered Apr 23 '21 at 10:43

1

you can use map -

a = {'education' : 1,'cinema' : 0}
train_docs['Class'] = train_docs['Class'].map(a)

answered Apr 23 '21 at 10:43

Nk03

14,699
2
8
22

score 0 · Accepted Answer · answered Apr 23 '21 at 10:48

When you use .apply method of pandas.Series given function should accept what that pandas.Series is holding, in this case strs,

def numconv(a):
    return a.map({'education' : 1,'cinema' : 0})

will not work if a is str, you might repair it by changing to

def numconv(a):
    return {'education' : 1,'cinema' : 0}[a]

or in this case just use .replace method of pandas.Series like so

import pandas as pd
df = pd.DataFrame({"Class":["education","education","education","cinema","cinema"]})
df["Class"].replace({'education' : 1,'cinema' : 0},inplace=True)
print(df)

output

Beware that function I proposed will fail if any other value will appears, whilst .replace ignores unknown values

Pandas Dataframe Column Value Conversion from string to binary

2 Answers2