Mode/Median/Mean of a 3d numpy array

Question

I have a 3d numpy array and my goal is to get the mean/mode/median of it.

It has a shape of [500,300,3]

And I would like to get for example:

[430,232,22] As the mode

Is there a way to do this? The standard np.mean(array) gives me a very large array.

I don't know if this is actually right?

weather_image.mean(axis=0).mean(axis=0)

It gives me a 1d np array with a length of 3

Answer 1

You want to get the mean/median/mode along the first two axes. This should work:

data = np.random.randint(1000, size=(500, 300, 3))

>>> np.mean(data, axis=(0, 1)) # in nunpy >= 1.7
array([ 499.06044   ,  499.01136   ,  498.60614667])
>>> np.mean(np.mean(data, axis=0), axis=0) # in numpy < 1.7
array([ 499.06044   ,  499.01136   ,  498.60614667])
>>> np.median(data.reshape(-1, 3), axis=0)
array([ 499.,  499.,  498.]) # mode
>>> np.argmax([np.bincount(x) for x in data.reshape(-1, 3).T], axis=1)
array([240, 519, 842], dtype=int64)

Note that np.median requires a flattened array, hence the reshape. And bincount only handles 1D inputs, hence the list comprehension, coupled with a little transposition magic for unpacking.