Store python array in each entry of a column

Question

I have got an array 'mutlilabel' which looks like this:

       [[0, 0, 0, 1, 0, 0, 0, 0],
       [1, 0, 0, 0, 0, 0, 0, 0],
       [1, 0, 0, 1, 0, 0, 0, 0],
                  ...
       [0, 0, 0, 1, 0, 0, 0, 0],
       [1, 0, 0, 0, 0, 0, 0, 0]]

and want to store each of those arrays in my target variable as I am facing a multi-label classification task. How can I achieve that? My code:

pd.DataFrame(multilabel)

Outputs multiple columns:

0   1   2   3   4   5   6   7

0   0   0   0   0   1   0   0   0
1   1   0   0   0   0   0   0   0
2   1   0   0   0   0   0   0   0

Thanks in advance!

Are you saying you just want the dataframe to have a single column with arrays of length 8 as each value in the column? — mdgrogan
– mdgrogan, Commented Sep 1, 2021 at 14:08
exactly! I want this as my target variable for a neural network — soulwreckedyouth
– soulwreckedyouth, Commented Sep 1, 2021 at 14:14

Niv Dudovitch · Accepted Answer · 2021-09-01 14:15:39Z

2

df = pd.DataFrame(list(multilabel))
list_column = df.apply(lambda row: row.values, axis=1)
pd.DataFrame(list_column, columns=['list_column'])

Result df:

answered Sep 1, 2021 at 14:15

Niv Dudovitch

1,6789 silver badges15 bronze badges

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

3 Comments

Niv Dudovitch Over a year ago

But you don't need to change your multilabel list of lists to a df to train your NN.

soulwreckedyouth Over a year ago

Can I just pass it? I am new to NN and especially to the Huggingface library with its Dataset objects. I have to do the train_test_split like splitting it into train_data and test_data and not the x_train/x_test/y_train/y_test structure. Thats why I want to transfer that multilabel list into a dataframe structure...

Niv Dudovitch Over a year ago

from sklearn.model_selection.train_test_split: "Allowed inputs are lists, numpy arrays, scipy-sparse matrices or pandas dataframes."

rpanai · Accepted Answer · 2021-09-01 16:35:59Z

1

Have you consider using the following trick?

import pandas as pd

arr =  [[0, 0, 0, 1, 0, 0, 0, 0],
       [1, 0, 0, 0, 0, 0, 0, 0],
       [1, 0, 0, 1, 0, 0, 0, 0],
       [0, 0, 0, 1, 0, 0, 0, 0],
       [1, 0, 0, 0, 0, 0, 0, 0]]

pd.DataFrame([arr]).T

Output

                          0
0  [0, 0, 0, 1, 0, 0, 0, 0]
1  [1, 0, 0, 0, 0, 0, 0, 0]
2  [1, 0, 0, 1, 0, 0, 0, 0]
3  [0, 0, 0, 1, 0, 0, 0, 0]
4  [1, 0, 0, 0, 0, 0, 0, 0]

EDIT In case you are using numpy arrays you can use the following

import numpy as np

pd.DataFrame(np.array(arr))\
  .apply(lambda x: np.array(x), axis=1)

edited Sep 1, 2021 at 16:35

answered Sep 1, 2021 at 14:08

rpanai

13.5k3 gold badges48 silver badges65 bronze badges

9 Comments

soulwreckedyouth Over a year ago

Doesn't work for me! It just transposes my 8x1026 array into a dataframe od 1024x8

Jiří Baum Over a year ago

That was @soulwreckedyouth, sorry

Jiří Baum Over a year ago

Probably the data is in a numpy array, rather than a python list of lists; that gives exactly this error

rpanai Over a year ago

Thank Jiri. @soulwreckedyouth as future reference try to ask question using these resources how-to-ask mcve. If in your question you use lists instead of numpy arrays people try to answer using the same format of data you provided.

soulwreckedyouth Over a year ago

sorry for that. will do that in the future

|

Jiří Baum · Accepted Answer · 2021-09-02 23:59:39Z

1

So, the real question is why... it doesn't seem like the most useful data structure.

That said, the one-dimensional data type in pandas is the Series:

>>> pd.Series(multilabel)
0    [0, 0, 0, 1, 0, 0, 0, 0]
1    [1, 0, 0, 0, 0, 0, 0, 0]
2    [1, 0, 0, 1, 0, 0, 0, 0]
3    [0, 0, 0, 1, 0, 0, 0, 0]
4    [1, 0, 0, 0, 0, 0, 0, 0]
dtype: object

You can then convert it further into a DataFrame:

>>> pd.DataFrame(pd.Series(multilabel))
                          0
0  [0, 0, 0, 1, 0, 0, 0, 0]
1  [1, 0, 0, 0, 0, 0, 0, 0]
2  [1, 0, 0, 1, 0, 0, 0, 0]
3  [0, 0, 0, 1, 0, 0, 0, 0]
4  [1, 0, 0, 0, 0, 0, 0, 0]

Edit: Per further discussion, this works if multilabel is a nested Python list, but not if it's a NumPy array.

edited Sep 2, 2021 at 23:59

answered Sep 1, 2021 at 14:17

Jiří Baum

6,9882 gold badges19 silver badges19 bronze badges

Collectives™ on Stack Overflow

Store python array in each entry of a column

3 Answers 3

3 Comments

9 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

3 Comments

9 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related