CODEWITHSUNDEEP
pythonscikit-learn
Alan
Alan

Reputation: 129

Multi-label encoding in scikit-learn

I am trying to encode the target column of my DataFrame. The type of the variable contained by this column is object.

I have a DataFrame that contains all the codes - icd10. Using those, I am trying to binarize the labels of my infoDF DataFrame.

My code looks like this:

from sklearn import preprocessing

lb = preprocessing.LabelBinarizer()
#fit all the possible label codes
lb.fit(icd10['ICD10'])
temp = lb.transform(infoDF['Target'])
for i,x in enumerate(lb.classes_):
    infoDF[x] = temp[ : , i]

When I run it, I get the following traceback:

  File "<ipython-input-42-2b1db450b16e>", line 3, in <module>
    lb.fit(icd10['ICD10'])

  File "C:\Users\as\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\preprocessing\label.py", line 413, in fit
    self.classes_ = unique_labels(y)

  File "C:\Users\as\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\utils\multiclass.py", line 96, in unique_labels
    raise ValueError("Unknown label type: %s" % repr(ys))

ValueError: Unknown label type: (0              1
1              2
2              3
3              4
4              5
5              6
6              7
7              8
8              9
9             10
10            11
11            12
12            13
13            14
14            15
15            16
16            17
17            18
18            19
19            20
20            21
21            22
22           A00
23          A000
24          A001
25          A009
26       A00–A09
27           A01
28          A010
29          A011

19433       Z960
19434       Z961
19435       Z962
19436       Z963
19437       Z964
19438       Z965
19439       Z966
19440       Z967
19441       Z968
19442       Z969
19443        Z97
19444       Z970
19445       Z971
19446       Z972
19447       Z973
19448       Z974
19449       Z975
19450       Z978
19451        Z98
19452       Z980
19453       Z981
19454       Z982
19455       Z988
19456        Z99
19457       Z990
19458       Z991
19459       Z992
19460       Z993
19461       Z998
19462       Z999
Name: ICD10, Length: 19463, dtype: object,)

I am not sure what I am doing wrong..

Upvotes: 3

Views: 1178

Answers (1)

Maximilian Peters
Maximilian Peters

Reputation: 31739

Although we don't have the exact format of your data set, it looks like the initial integers cause the problem.

sklearn's LabelBinarizer calls sklearn.utils.multiclass.unique_labels which according to the documentation does not allow "a mix of string and integer labels".

Try to remove the first 21 rows and see if the error persists.

import pandas as pd
from sklearn import preprocessing

lb = preprocessing.LabelBinarizer()

icd11 = pd.DataFrame({'ICD11': [0, '6C51', '6C50.Z']})

# crashes
lb.fit(icd11['ICD11'])

# does not crash
lb.fit(icd11['ICD11'][1:])

Upvotes: 4

Related Questions