In the example, the map method shows to take a lambda with two parameters (key and xy), but it appears as though the python version of spark only has a map method that expects a lambda with just a single parameter.
So instead of the following
r = sumCount.map(lambda key, xy: (key, xy[0]/xy[1])).collectAsMap()
We should use
r = sumCount.map( lambda kvp: ( kvp[0], kvp[1][0] / kvp[1][1] ) ).collectAsMap()