-1

I have a dataframe with 10 columns with each elements as list of length 40.How can i convert them into numpy array of shape (-1,400) . column (click this link to see the column)

SeaBean
  • 22,547
  • 3
  • 13
  • 25
Nandhagopal cs
  • 11
  • 1
  • Please add a [Minimal, Reproducible Example](https://stackoverflow.com/help/minimal-reproducible-example) and don't only provide links to images. In particular, the dtype of the dataframe is important here. – Jan Wilamowski Jun 17 '21 at 06:21
  • 1
    Look at `df.to_numpy()`. That's the array created from the frame. A clear idea of it's shape, dtype, and nature of the elements is required before you can consolidate into a more compact array. – hpaulj Jun 17 '21 at 06:27
  • Array shape of (-1,400)? What does it mean? – sharathnatraj Jun 17 '21 at 06:32
  • This question has already been asked. Check this out: https://stackoverflow.com/questions/31789160/convert-select-columns-in-pandas-dataframe-to-numpy-array. Hopefully this answers your question. –  Jun 17 '21 at 06:43
  • @desertnaut The quoted link does not completely solve this question. Need to concatenate the lists within columns. – SeaBean Jun 17 '21 at 10:14
  • 1
    @SeaBean OK, I reopened it – desertnaut Jun 17 '21 at 10:18

1 Answers1

1

Try: df.apply() + numpy.concatenate() + to_numpy():

import numpy as np

df.apply(lambda x: np.concatenate(x), axis=1).to_numpy()

Test Run

Define a dataframe with 3 rows 10 columns each element of a list of length 40:

data = [[list(range(40))] * 10] * 3
cols = list(range(10))
df = pd.DataFrame(data, columns=cols)

Result:

Shape of whole extraction:

df.apply(lambda x: np.concatenate(x), axis=1).to_numpy().shape

(3,)

Shape of each row:

df.apply(lambda x: np.concatenate(x), axis=1).to_numpy()[0].shape

(400,)
SeaBean
  • 22,547
  • 3
  • 13
  • 25