Python list vector matrix decimals 0s 1s

Question

I have a list in python

l=[[0.1,0.2,0.9],[0.3,0.4,0.6],[0.8,0.2,0.8]]

if element <=0.5, return 0, if element > 0.5, return 1

Hence resulting l should look like:

l=[[0,0,1],[0,0,1],[1,0,1]]

Thanks in advance!

What is your question? This is not a code-writing service. It is a question and answer site. — juanpa.arrivillaga, Apr 28 '18 at 19:00
Take a look at [list comprehensions](http://www.secnetix.de/olli/Python/list_comprehensions.hawk). — Patrick Haugh, Apr 28 '18 at 19:01

bla · Answer 1 · 2018-04-28T21:07:53.100

0

Check out round.

>>> l=[[0.1,0.2,0.9],[0.3,0.4,0.6],[0.8,0.2,0.8]]
>>> [[round(x) for x in i] for i in l]
[[0, 0, 1], [0, 0, 1], [1, 0, 1]]

As mentioned by @iAdjunct in the comments and discussed in this question, if working on python 2 this will not result in desired output.

That is because they adopt different rounding methods. The difference can be illustrate with these examples:

Python2:

>>> round(0.5)
1.0

Python3

>>> round(0.5)
0

edited Apr 28 '18 at 21:07

answered Apr 28 '18 at 19:47

bla

1,840
1
13
17

Note: his function is actually not the same as round because, for some reason, he's rounding 0.5 *down* instead of *up* – iAdjunct Apr 28 '18 at 19:53
Just tested in python3.6 and `round(0.5)` returns `0`. – bla Apr 28 '18 at 19:59
1

`Python 2.7.10` returns `1.0`. That is weird; Python2's `round` function is consistent with the traditional definition of round, but Python3's implements the round-half-to-even paradigm (https://en.wikipedia.org/wiki/Rounding#Round_half_to_even). – iAdjunct Apr 28 '18 at 20:25

score 0 · Answer 2 · answered Apr 28 '18 at 21:22

There are couple of ways you can do this; either in pure Python via list comprehension or in a vectorised fashion using a 3rd party library.

Pure Python

You can use a nested list comprehension.

l = [[0.1,0.2,0.9],[0.3,0.4,0.6],[0.8,0.2,0.8]]

res = [[0 if i <= 0.5 else 1 for i in row] for row in l]

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

Vectorised alternative

This uses the 3rd party library numpy.

import numpy as np

res = np.where(np.array(l) <= 0.5, 0, 1)

array([[0, 0, 1],
       [0, 0, 1],
       [1, 0, 1]])

Note on rounding

Comparing floats with floats is difficult due to the nature of float not storing numeric data in base-10 format. See Is floating point math broken? for more details. This is a problem even with the above methods.

This means you may have to take particular care for boundary cases, if this matters to you.

Python list vector matrix decimals 0s 1s

2 Answers2