How can I find maximum along multiple axes of a multidimensional numpy array?

Question

I have a numpy array of these dimensions

data.shape (categories, models, types, events, days) -> (10, 11, 50, 100, 14)

Now, I want to find the maximum of the 14 days for all events for each of the 11 models. But I am not sure how to do it in the numpy way. I am not sure if this is correct.

modelmax = []
nmodels = 11

for modelcount in range(nmodels):
    modelmax.append(np.max(data[0][modelcount][:], axis=2))

As an example, for the 100 events:

np.max(data, axis=4)[0][0][0])

[ 3.9264417   3.3029506   3.0707457   3.6646023   1.7508441   3.1634364
  6.195052    1.5353022   1.8033538   1.4508389   1.3882699   2.0849068
  3.654939    6.6364765   3.92829     6.6467876   1.5442419   4.639682
  9.361191    5.261462    1.7438816   5.6970205   2.4356377   1.6073244
  2.6177561   6.886767    3.890399    2.8880894   1.9826577   1.0888597
  4.3763924   3.8597727   1.790302    1.0277777   6.270729    9.311213
  2.318774    2.9298437   1.139397    0.9598383   3.0489902   1.6736581
  1.3983868   2.0979824   4.169757    1.0739225   1.5311266   1.4676268
  1.726325    1.8057758   2.226462    2.6197987   4.49518     2.3042605
  5.7164993   1.182242    1.5107205   2.2920077   2.205539    1.4702082
  2.154468    2.0641963   4.9628353   1.9987459   2.1360166   1.7073958
  1.943267    7.5767093   1.3124634   2.2648168   1.1504744   3.210688
  2.6720855   2.998225    4.365262    3.5410352  10.765423    4.6292825
  3.1789696   0.92157686  1.663245    1.5835482   3.1070056   1.6918416
  8.086268    3.7994847   2.4314868   1.6471033   1.1688241   1.7820593
  3.3509188   1.3092748   3.7915008   1.018912    3.2404447   1.596657
  2.0869658   2.6753283   2.1096318   8.786542  ]

I have also tried

np.max(dryflow[0][:], axis=3)

But these multidimensional indices are leaving me confused.

@Chiel I am trying to reduce the complexity of the problem for understanding purposes. But if I could have the daily maximums for category, model, type and event, that would be swell....I am just very confused regarding the multi-dims, and I am probably not relaying what I don't understand... — maximusdooku
– maximusdooku, Commented Nov 7, 2018 at 22:16
Eventually, I will take the log of the 14 day maximums and find the overall maximum and minimum for every type...Hope it explains a bit.. — maximusdooku
– maximusdooku, Commented Nov 7, 2018 at 22:18
In numpy it is better to use [0,0,0,0] style of indexing rather than [0][0][0][0]. Sometimes they produce the same thing, but sometimes the differences give problems. Also with arrays [:] does nothing for you. — hpaulj
– hpaulj, Commented Nov 7, 2018 at 22:20
@hpaulj Thanks! I didn't know there was any difference. BTW, can you expand a little bit on [:] doing nothing.. — maximusdooku
– maximusdooku, Commented Nov 7, 2018 at 22:23

Tomi Aarnio · Accepted Answer · 2019-12-17 11:35:40Z

10

If you want the maximum over days for each combination of (category, model, type, event), for a resulting shape of (10, 11, 50, 100, 1):

np.max(data, axis=4)

If you want the maximum over days and types for each combination of (category, model, event), for a resulting shape of (10, 11, 100, 1):

np.max(data, axis=(2,4))

answered Dec 17, 2019 at 11:35

Tomi Aarnio

2,5571 gold badge21 silver badges16 bronze badges

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

Collectives™ on Stack Overflow

How can I find maximum along multiple axes of a multidimensional numpy array?

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related