return numpy.flipud(numpy.array(arr, float))

I guess this one can be solved in a lot of different ways... list.reversed() before transforming it to array, or using [::-1] to sort. But feeling like the idea of this "challange" is to get common with "numpy" module and to use all of it's capabilities. Even tho this is not relevant in this particular task and given test cases but using flipud would be the best approach from learning perspective. However from some reserches done by other people () - either slicing [::-1]or flipud gives almost the same performance results bigger the array becomes. But defenitely outperforms list.reverse()