Parameters —
 arr: [array_like] input
 axis : axis [int or tuples of int] along which we want to get the maximum value. Otherwise, it will assume that arr will be flattened.
 from: [ndarray, optional] an alternative output array where to put the result
 keepdmis: [boolean, optional] if this parameter is set to True, the downsized axes remain in
the result as dimensions with size one. With this option, the result will translate correctly
the input array. If a default is passed, then keepdims will not be passed to all
ndarray subclasses, however any nondefault will be. If the subclass sum method
does not implement keepdims, exceptions will be thrown.
Return — Maximum array — arr [ndarray or scalar], scalar if axis is None; The result is an array of dimensions a.ndim — 1 if axis is specified.
Code —

Exit —
arr: [0 1 2 3 4 5 6 7] Max of arr: 7 arr: [[0 1 2 3 4] [5 6 7 8 9]] Max of arr, axis = None: 9 Max of arr, axis = 0: [5 6 7 8 9] Max of arr, axis = 1: [4 9]
Link —
https://docs.scipy.org/doc/numpy1.13.0/reference/generated/numpy.amax.html
Note —
These codes will not work for online IDs. Please run them on your systems to see how they work
This article is courtesy of Mohit Gupta_OMG