Create a DataFrame for MultiClass classification

Question

I have a data frame with n rows, I want to assign a class to every row randomly from m classes such that the proportion of all classes are the same.

Example:

>>> classes = ['c1','c2','c3','c4']
>>> df = pd.DataFrame(np.random.randn(100, 5), columns = list("abcde"))
>>> df
           a         b         c         d         e
0  -0.341559  1.499159  0.269614 -0.198663 -1.081290
1  -1.966477  1.902292 -0.092296 -1.730710 -1.342866
2   1.188634 -2.851902  1.130480 -0.495677 -0.569557
3  -0.816190  1.205463  1.157507 -0.217025 -0.160752
4  -2.001114 -0.818852 -0.696057 -0.874615 -0.577101
..       ...       ...       ...       ...       ...
95  0.502192  0.434275  0.358244 -0.763562 -0.787102
96 -1.071011  0.045387  0.297905 -0.120974  0.185418
97  2.458274 -1.852953 -0.049336 -0.150604 -0.292824
98  1.992513 -0.431639  0.566920 -1.289439  0.626914
99  0.685915 -0.723009 -0.168497  1.630057  1.587378

[100 rows x 5 columns]

Expected output:

>>> df
           a         b         c         d         e class
0  -0.341559  1.499159  0.269614 -0.198663 -1.081290    c3
1  -1.966477  1.902292 -0.092296 -1.730710 -1.342866    c4
2   1.188634 -2.851902  1.130480 -0.495677 -0.569557    c2
3  -0.816190  1.205463  1.157507 -0.217025 -0.160752    c3
4  -2.001114 -0.818852 -0.696057 -0.874615 -0.577101    c1
..       ...       ...       ...       ...       ...   ...
95  0.502192  0.434275  0.358244 -0.763562 -0.787102    c1
96 -1.071011  0.045387  0.297905 -0.120974  0.185418    c3
97  2.458274 -1.852953 -0.049336 -0.150604 -0.292824    c2
98  1.992513 -0.431639  0.566920 -1.289439  0.626914    c1
99  0.685915 -0.723009 -0.168497  1.630057  1.587378    c2

[100 rows x 6 columns]

With the class proportions being the same

@Wanderer yes, that is how the above output was created but this does not ensure equal class proportions. — darth baba
– darth baba, Commented Jun 22, 2021 at 11:29
@dathbaba actually it does, if you set 'weights' equally then probability of selecting them will be the same = equal class proportions — Wanderer
– Wanderer, Commented Jun 22, 2021 at 11:31
Nope the class distribution is not same, df.groupby('class').size() confirms it. equal class proportions means there are 25 rows of each class — darth baba
– darth baba, Commented Jun 22, 2021 at 14:42

pzmn · Accepted Answer · 2021-06-23 18:54:18Z

0

This should do the job

classes = ['c1','c2','c3','c4']
df = pd.DataFrame(np.random.randn(100, 5), columns = list("abcde"))

classes = np.repeat(classes, df.shape[0]/len(classes))
np.random.shuffle(classes)
df['class'] = classes

answered Jun 23, 2021 at 18:54

pzmn

1545 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Create a DataFrame for MultiClass classification

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related