Scikit-learn不适用于KNN上的字符串值

Question

I'm using Scikit learn to do a K-Nearest Neigbour Classification:

from sklearn.neighbors import KNeighborsClassifier

model=KNeighborsClassifier() 
model.fit(train_input,train_labels)

If I print my data:

print("train_input:")
print(train_input.iloc[0])
print("\n")
print("train_labels:")
print(train_labels.iloc[0]) 

I get this:

train_input:
PassengerId                          1
Pclass                               3
Name           Braund, Mr. Owen Harris
Sex                               male
Age                                 22
SibSp                                1
Parch                                0
Ticket                       A/5 21171
Fare                              7.25
Cabin                              NaN
Embarked                             S
Name: 0, dtype: object


train_labels:
0

The code fails with this error:

ValueError                                Traceback (most recent call last)
<ipython-input-21-1f18eec1e602> in <module>()
     63 
     64 model=KNeighborsClassifier()
---> 65 model.fit(train_input,train_labels)
ValueError: could not convert string to float: 'Q'

So, does the KNN algorithm not work with String values?

How can I modify my data such that it fits the KNN implementation in Scikit-Learn?

Answer 1

For nominal String features, consider one hot encoding: http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.OneHotEncoder.html .

For ordinal String features, consider label encoding (with a sensible ordering based on your understanding of the feature): http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.LabelEncoder.html .