create numpy array with varying shape

Question

I want to create a numpy array in order to fill it with numpy arrays. for example:

a = [] (simple array or numpy array) 
b = np.array([[5,3],[7,9],[3,8],[2,1]])
a = np.concatenate([a,b])
c = np.array([[1,2],[2,9],[3,0]])
a = np.concatenate([a,c])

I would like to do so because I have wav files from which I extract some features so I can't read from 2 files concurrently but iteratively. How can I create an empty ndarray with the second dimension fixed e.g. a.shape = (x,2) or how can I concatenate the arrays even without the creation of a "storage" array ?

Trying to call concatenate incrementally is really, really slow. It's better to build up a list of arrays to concatenate and then concatenate them all at once. — user2357112
– user2357112, Commented May 4, 2017 at 17:08
@user2357112 yes. I realized that myself. check my answer. thanks — stefos
– stefos, Commented May 4, 2017 at 17:34

stefos · Accepted Answer · 2017-05-04 17:31:04Z

1

Actually there are 2 options. The first one is: a = np.empty((0, 2)) , which creates an empty np array with the first dimension varying. The second is to create an empty array a = [] , append the np arrays in the array and then use np.vstack to concatenate them all together in the end. The latter the most efficient option.

answered May 4, 2017 at 17:31

stefos

237 bronze badges

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

Comments

Nuageux · Accepted Answer · 2017-05-04 14:31:33Z

0

You have to had brackets in concatenate function:

b = np.array([[5,3],[7,9],[3,8],[2,1]])
c = np.array([[1,2],[2,9],[3,0]])
a = np.concatenate([b,c])

Output:

[[5 3]
 [7 9]
 [3 8]
 [2 1]
 [1 2]
 [2 9]
 [3 0]]

answered May 4, 2017 at 14:31

Nuageux

1,6861 gold badge18 silver badges29 bronze badges

Collectives™ on Stack Overflow

create numpy array with varying shape

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related