Fill array with rows of different lenghts Python

Question

So I found this: Numpy: Fix array with rows of different lengths by filling the empty elements with zeros

But what I actually want is this:

mylist = [[1],[1,2],[1,2,3]]

mylist.fill()
>>> [[0,0,1], [0,1,2], [1,2,3]]

I know that pandas' fillna fills but the 0 are at the right part of my matrix and I need them at the left part. Any clues?

From the Help center: Check the spelling of your question carefully. — user202729, Jan 30 '18 at 15:45

score 4 · Accepted Answer · answered Jan 30 '18 at 15:47

4

I think this should do it:

def fill(a):
    length = max([len(i) for i in a])
    return [[0]*(length-len(i)) + i for i in a]

fill(mylist)
#[[0,0,1], [0,1,2], [1,2,3]]

answered Jan 30 '18 at 15:47

zipa

27,316
6
40
58

score 2 · Answer 2 · answered Jan 30 '18 at 15:48

2

Since you tag pandas

pd.DataFrame(mylist).\
  apply(lambda x: sorted(x, key=pd.notnull), 1).\
    fillna(0).astype(int).values.tolist()
Out[89]: [[0, 0, 1], [0, 1, 2], [1, 2, 3]]

answered Jan 30 '18 at 15:48

BENY

317,841
20
164
234

This looks interesting. It would be great if you can add some more details how this works – Sohaib Farooqi Jan 30 '18 at 15:55
@GarbageCollector using constructor creat data frame, sort each row by using NaN or not (If NaN move to the front), then convert back to list – BENY Jan 30 '18 at 15:56
@GarbageCollector its simple sorting of row elements using the key. – Bharath M Shetty Jan 30 '18 at 15:57

Bharath M Shetty · Answer 3 · 2018-01-30T15:55:46.013

1

Fillna with 0 and sort values check if they are not 0's i.e

df = pd.DataFrame(mylist)
df.fillna(0).apply(lambda x : sorted(x,key=lambda x : x!=0),1).values.astype(int).tolist()

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

edited Jan 30 '18 at 15:55

answered Jan 30 '18 at 15:53

Bharath M Shetty

30,075
6
57
108

1

Maybe add astype( int ) :-) – BENY Jan 30 '18 at 15:54

Fill array with rows of different lenghts Python

3 Answers3