NumPy array element not getting updated

Question

I have a NumPy array as follows:

supp = np.array([['A', '5', '0'], ['B', '3', '0'], ['C', '4', '0'], ['D', '1', '0'], ['E', '2', '0']])

Now, I want to update the row[2] as row[1]/6. I'm using..

for row in supp: row[2] = row[1].astype(int) / 6

But row[2] seems to remain unaffected..

>>> supp
array([['A', '5', '0'],
   ['B', '3', '0'],
   ['C', '4', '0'],
   ['D', '1', '0'],
   ['E', '2', '0']],  
  dtype='<U1')

I'm using Python 3.5.2 and NumPy 1.11.1.

Any help is appreciated. Thanks in advance

Trying to put strings and numbers in the same array is a bad idea. Depending on what you're doing, Pandas might have more suitable tools for your use case, or it might be better to just get rid of the first column and use an array of float dtype. — user2357112
– user2357112, Commented Aug 19, 2016 at 3:02

Julien · Accepted Answer · 2016-08-19 03:21:33Z

5

The problem is that an np.array has only one type which is automatically assumed to be strings supp.dtype == '|S1' since your input contains only strings of length 1. So numpy will automatically convert your updated inputs to strings of length 1, '0's in your case. Force it to be of generic type object and then it will be able to have both strings and ints or floats or anything else:

supp = np.array([['A', '5', '0'], ['B', '3', '0'], ['C', '4', '0'], ['D', '1', '0'], ['E', '2', '0']])
supp = supp.astype(object)

for row in supp:
    row[2] = int(row[1]) / 6

result:

[['A' '5' 0.8333333333333334]
 ['B' '3' 0.5]
 ['C' '4' 0.6666666666666666]
 ['D' '1' 0.16666666666666666]
 ['E' '2' 0.3333333333333333]]

alternatively you can also use the dtype '|Sn' with larger value of n:

supp = np.array([['A', '5', '0'], ['B', '3', '0'], ['C', '4', '0'], ['D', '1', '0'], ['E', '2', '0']])
supp = supp.astype('|S5')

for row in supp:
    row[2] = int(row[1]) / 6

result:

[['A' '5' '0.833']
 ['B' '3' '0.5']
 ['C' '4' '0.666']
 ['D' '1' '0.166']
 ['E' '2' '0.333']]

and in this case you are still having only strings if that is what you want.

edited Aug 19, 2016 at 3:21

answered Aug 19, 2016 at 2:55

Julien

15.3k6 gold badges33 silver badges58 bronze badges

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

2 Comments

lu5er Over a year ago

Thanks. Learned a new thing today. :)

Julien Over a year ago

Me too! ;) BTW, I totally second user2357112 's comment. You should probably think about it... Using dtype=object might be convenient but you'll loose a lot of speedup on numerical computations, which is usually the point of using numpy.

Collectives™ on Stack Overflow

NumPy array element not getting updated

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related