Sort column having float values in python

Question

This is my list :

vec = [[(0, 6.369426751591834e-05),
      (1, 6.369426751591834e-05),
      (2, 6.369426751591834e-05),
      (3, 6.369426751591834e-05),
      (4, 6.369426751591834e-05),
      (5, 6.369426751591834e-05),
      (6, 6.369426751591834e-05),
      (7, 0.0007714083510261222),
      (8, 6.369426751591834e-05),  
      ...........................
      (4995, 6.369426751591834e-05),
      (4996, 6.369426751591834e-05),
      (4997, 6.369426751591834e-05),
      (4998, 6.369426751591834e-05),
      (4999, 6.369426751591834e-05)]]

I want to sort it based 2nd column which has float value. Plus how to print first 10 from the output list. I tried many methods, but couldn't figure out. Any suggestions?

Paulo, it was output of lda algorithm, it was nested by default. — sb32134
– sb32134, Commented Mar 25, 2014 at 15:15

drez90 · Accepted Answer · 2014-03-25 15:19:08Z

3

sorted(vec[0], key=lambda x: x[1])

vec[0] only because you have a nested list... normally it would just be vec

edited Mar 25, 2014 at 15:19

answered Mar 25, 2014 at 15:13

drez90

8145 silver badges17 bronze badges

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

2 Comments

jonrsharpe Over a year ago

This is a TypeError; key must be a keyword argument

sb32134 Over a year ago

Thanks for the answer, but above one by @Grijesh worked. I tried yours, it gives :TypeError: <lambda>() takes exactly 1 argument (2 given)

jonrsharpe · Accepted Answer · 2014-03-25 15:12:34Z

2

Generally, you can sort by the ith value in a container using operator.itemgetter(i) as the key for sorting:

from operator import itemgetter

vec[0] = sorted(vec[0], key=itemgetter(1))

On your example, with some duplicates removed:

>>> vec = [[(0, 6.369426751591834e-05),
            (7, 0.0007714083510261222),
            (8, 6.369426751591834e-05)]]
>>> sorted(vec[0], key=itemgetter(1))
[(0, 6.369426751591834e-05), 
 (8, 6.369426751591834e-05), 
 (7, 0.0007714083510261222)]

answered Mar 25, 2014 at 15:12

jonrsharpe

123k31 gold badges277 silver badges488 bronze badges

2 Comments

Simon Over a year ago

A little overkill. @d_rez90's is simpler

sb32134 Over a year ago

This one worked too! Thanks. I can't upvote your replies guys, i don't have enough points.

CT Zhu · Accepted Answer · 2014-03-25 15:25:34Z

2

numpy way of doing it:

In [7]:
AV=np.array(vec)
AV[:,np.argsort(AV[:,:,1])]
Out[7]:
array([[[[  0.00000000e+00,   6.36942675e-05],
         [  1.00000000e+00,   6.36942675e-05],
         [  2.00000000e+00,   6.36942675e-05],
         [  3.00000000e+00,   6.36942675e-05],
         [  4.00000000e+00,   6.36942675e-05],
         [  5.00000000e+00,   6.36942675e-05],
         [  6.00000000e+00,   6.36942675e-05],
         [  8.00000000e+00,   6.36942675e-05],
         [  4.99500000e+03,   6.36942675e-05],
         [  4.99600000e+03,   6.36942675e-05],
         [  4.99700000e+03,   6.36942675e-05],
         [  4.99800000e+03,   6.36942675e-05],
         [  4.99900000e+03,   6.36942675e-05],
         [  7.00000000e+00,   7.71408351e-04]]]])

There is a np.set_printoptions() for controlling print options. There is no way, however, to only print the first 10 lines, you can instead print the first n lines and the last n lines by: np.set_printoptions(threshold=a_samll_number, edgeitems=n)

edited Mar 25, 2014 at 15:25

answered Mar 25, 2014 at 15:14

CT Zhu

54.6k18 gold badges125 silver badges136 bronze badges

1 Comment

sb32134 Over a year ago

Thanks, that printing option is really handy for observing first and last ten results.

Simon · Accepted Answer · 2014-03-25 15:19:03Z

1

As you asked to print the first ten:

srtd = sorted(vec[0], lambda x: x[1])
print srtd[:10]

answered Mar 25, 2014 at 15:19

Simon

2,9803 gold badges20 silver badges26 bronze badges

6 Comments

Simon Over a year ago

Put a minus in front of x[1] if you want to flip sort order.

sb32134 Over a year ago

I already tried vec[:10], it prints the whole list in my case

Simon Over a year ago

That's because your list is nested. Above will work as it takes out the first element. You would need vec[0][:10]. Or simply don't nest the list inside another list.

sb32134 Over a year ago

yes, it works after adding key: as pointed above: sorted(vec[0], key=lambda x: x[1])

Simon Over a year ago

When you declare it, you are declaring it with two square brackets. Use one. Then you can remove the [0] from a lot of the solutions. The numpy one works well as he is extracting the inner list first with: [:,:,1]. However, that seems a little overkill. My modification of @d_rev90's is the simplest approach.

|

Collectives™ on Stack Overflow

Sort column having float values in python

4 Answers 4

2 Comments

2 Comments

1 Comment

6 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

2 Comments

2 Comments

1 Comment

6 Comments

Your Answer

Sign up or log in

Post as a guest

Related