Axis play a key role in Numpy array operations, it indicate the dimension of an array(or a tensor), for example, if an array have shape (2,4,6),

the first dimension is axis=0 coresponding to shape 2, second dimension is axis=1 coresponding to shape 4, etc, and for the convience, Numpy also can use axis=-1 to indicate the last dimension.

I made an animation for intuiation.

Functions in Numpy to operate array can be classified by how to change dimensions.

  1. keep dimensions
  2. collapse dimensions
  3. expand dimensions
  4. switch axis

Here is a animation shows how to operate array along different axis when keep/collapse dimensions.

s2

For expand and sitch axis it is almost the same idea. Numpy doc example expain this very well.

   
expand_dims(a, axis) Expand the shape of an array.
swapaxes(a, axis1, axis2) Interchange two axes of an array.