get the argmax from an array and save it - python

Question

I have an array like the one below. How can I take the argmax from this array and save it as a dataframe.

array

([[  4.33578761e-03,   1.11002535e-01,   1.18002132e-01,
      2.23158062e-01,   3.57010156e-01,   9.33912754e-01],
   [  3.21856714e-05,   3.46616306e-03,   4.58798464e-03,
      7.65424687e-03,   1.02789648e-01,   9.99610841e-01]], dtype=float32) was passed

Wazy · Answer 1 · 2017-12-21T08:40:36.607

2

Considering n as numpy array

import numpy as np
n = np.array([[  4.33578761e-03,   1.11002535e-01,   1.18002132e-01,
      2.23158062e-01,   3.57010156e-01,   9.33912754e-01],
   [  3.21856714e-05,   3.46616306e-03,   4.58798464e-03,
      7.65424687e-03,   1.02789648e-01,   9.99610841e-01]], dtype=float)

# printing argmax at axis = 0
print(np.argmax(n, axis=0))
#print(n.argmax(axis=0))

a_max = np.argmax(n, axis=0)
np.save(open("your_file_path", "w"), a_max)

and to load

a_max = np.load(open("your_file_path"))

References

edited Dec 21 '17 at 08:40

answered Dec 20 '17 at 06:20

Wazy

8,822
10
53
98

can you save the print? – Reese Fitzmaurice Dec 20 '17 at 06:54
if you meant saving result of `argmax` in a variable, then yes you can. – Wazy Dec 20 '17 at 07:13
how can you do this? @Wazzzy – Reese Fitzmaurice Dec 20 '17 at 15:55

jezrael · Answer 2 · 2017-12-20T06:12:14.990

I believe you need:

print (arr.argmax(axis=1))
[5 5]

print (arr.argmax(axis=0))
[0 0 0 0 0 1]

First get argmax and then use DataFrame constructor:

print (pd.DataFrame(arr.argmax(axis=0), columns=['a']))
   a
0  0
1  0
2  0
3  0
4  0
5  1

print (pd.DataFrame(arr.argmax(axis=0)[None, :]))
   0  1  2  3  4  5
0  0  0  0  0  0  1

print (pd.DataFrame(arr.argmax(axis=1), columns=['a']))
   a
0  5
1  5

Or create DataFrame and call idxmax

print (pd.DataFrame(arr).idxmax().to_frame('a'))
#same as
#print (pd.DataFrame(arr).idxmax(axis=0).to_frame('a'))
   a
0  0
1  0
2  0
3  0
4  0
5  1

print (pd.DataFrame(arr).idxmax().to_frame(0).T)
   0  1  2  3  4  5
0  0  0  0  0  0  1


print (pd.DataFrame(arr).idxmax(axis=1).to_frame('a'))
   a
0  5
1  5

get the argmax from an array and save it - python

2 Answers2