If you know how to select subsets of Python lists, you know most of what you need to know about ndarray slicing. The elements of the array being indexed that correspond to the elements of the indexing object are returned in a new array. The most important aspect of indexing is to remember that there is more than one dimension, and the indexing method should be able to handle these other dimensions.
Remember the following points while selecting elements explicitly:
Separate the indexing objects for different dimensions with a comma; the object before the first comma shows how the first dimension is indexed. After the first comma comes the index for the second dimension, after the second comma comes the index for the third dimension, and so on.