Pandas不会将分类数据[性别]更改为数值[0/1]

Question

I am trying to work through the Titanic dataset. I want to convert the Sex column to binary values. This is my attempt:

sex = train_dataset['Sex'].replace([0,1],['female','male'],inplace=True)

And when I try to print(sex) , the console outputs None !

I have tried to implement other approaches on SO as well but none of them seem to work. Below is my full source code:

import pandas as pd
from numpy import corrcoef

train_dataset = pd.read_csv("https://raw.githubusercontent.com/oo92/Titanic-Kaggle/master/train.csv")
test_dataset = pd.read_csv("https://raw.githubusercontent.com/oo92/Titanic-Kaggle/master/test.csv")

survived = train_dataset['Survived']
pClass = train_dataset['Pclass']

#Doesn't work
sex = train_dataset['Sex'].replace([0,1],['female','male'],inplace=True)

age = train_dataset['Age']
fare = train_dataset['Fare']
parch = train_dataset['Parch']
sibSp = train_dataset['SibSp']

# print("Correlation between parent-children & survival is: " + str(corrcoef(survived, parch)))
# print("Correlation between age & survival is: " + str(corrcoef(survived, age)))
# print("Correlation between Siblings/Spouse & survival is: " + str(corrcoef(survived, sibSp)))

print(sex)

Answer 1

Try:

sex = train_dataset['Sex'].replace(['female','male'],[0,1])
print(sex)

It looks like your syntax is off. See the replace function

Output:

Answer 2

你可以使用np.where

dataset['sex']=np.where(dataset['sex']=='female',0,1)

Answer 3

Official documentation for the parameters:

inplace : bool, default False If True, in place. Note: this will modify any other views on this object (eg a column from a DataFrame). Returns the caller if this is True.

To summarize, inplace=True returns None and inplace=False returns a copy of the object with the operation performed.

So, in your case as the operation is inplace=True , the original series object train_dataset['Sex'] is modified. Try to print the train_dataset after the operation, you should see the modified dataframe.

Refer to the official documentation here

Answer 4

There are two problems here, first, you have turned around the arguments in .replace(<replace_this>, <with_this>) . Secondly, you are using the option inplace=True . This changes the train_dataset instance, instead of returning a value.

Now that you know that no value is returned when using inplace=True , you will understand that sex should be equal to None , because nothing is returned:

>>> import pandas as pd
>>> df = pd.DataFrame({'a': ['male', 'female', 'female', 'male']})
>>> df
        a
# 0    male
# 1  female
# 2  female
# 3    male

Without replace=True :

Now when we replace the values, we'd get

>>> df.replace(['female', 'male'], [0,1])
# a
# 0  1
# 1  0
# 2  0
# 3  1

But df itself still looks exactly the same as it did before:

>>> df
        a
#  0    male
#  1  female
#  2  female
#  3    male

So in order to replace the value in df , we would do:

>>> df['a'] = df['a'].replace(['male', 'female'], [0,1])
>>> df
#    a
# 0  0
# 1  1
# 2  1
# 3  0

With replace=True :

When you run this instead: df.replace(['female', 'male'], [0, 1], inplace=True)) , you would get a manipulated version of df back right away:

>>> df.replace(['female', 'male'], [0, 1], inplace=True)
>>> df
#    a
# 0  0
# 1  1
# 2  1
# 3  0

Note that inplace=True argument makes that no value is returned:

>>> test = df.replace(['female', 'male'], [0, 1], inplace=True)
>>> type(test)
# <class 'NoneType'>