python pandas\numpy encode unique by integers

Question

Say I have x=["apple","orange","orange","apple","pear"] I would like to have a categorical representation with integers e.g. y=[1,2,2,1,3]. What would be the best way to do so?

Presumably you checked the [docs](http://pandas.pydata.org/pandas-docs/stable/categorical.html)? — EdChum, Jan 14 '16 at 12:18
If you're working with `numpy` you can simply `np.unique(["apple","orange","orange","apple","pear"], return_inverse=True)[1]`, without turning to pandas — Sergey Bushmanov, Jan 14 '16 at 12:30

score 1 · Accepted Answer · answered Jan 14 '16 at 12:17

1

You could use pd.factorize and use field 0 for that:

In [465]: pd.factorize(x)
Out[465]: (array([0, 1, 1, 0, 2]), array(['apple', 'orange', 'pear'], dtype=object))

In [466]: pd.factorize(x)[0] + 1
Out[466]: array([1, 2, 2, 1, 3])

answered Jan 14 '16 at 12:17

Anton Protopopov

30,354
12
88
93

score 0 · Answer 2 · answered Jan 14 '16 at 12:17

You can use:

import pandas as pd

x=["apple","orange","orange","apple","pear"]
s = pd.Series(x)

print s

0     apple
1    orange
2    orange
3     apple
4      pear

print pd.Categorical(s).codes

[0 1 1 0 2]

Or:

import pandas as pd

x=["apple","orange","orange","apple","pear"]

print pd.Categorical(x).codes

#[0 1 1 0 2]

score -1 · Answer 3 · answered Jan 14 '16 at 12:16

-1

With Pandas: x.astype('category').cat.codes

answered Jan 14 '16 at 12:16

Hanan Shteingart

8,480
10
53
66

python pandas\numpy encode unique by integers

3 Answers3