Real-time Machine Learning with Spark Streaming - Machine Learning with Spark - page 388

(Miguel,(3,20.47))
(Saul,(1,5.49))
(Manuela,(2,18.939999999999998))
(Eric,(2,18.939999999999998))
...
-------------------------------------------
Time: 1416080441000 ms
-------------------------------------------
(Janet,(6,34.94))
(Juan,(4,33.92))
(Frank,(2,14.44))
(James,(7,48.93000000000001))
(Malinda,(1,9.99))
(Elaine,(7,61.89))
(Gary,(4,28.46))
(Michael,(1,8.95))
(Richard,(2,16.439999999999998))
(Miguel,(5,35.95))
...
We can see that the number of purchases and revenue totals for each user are added to
with each batch of data.
Tip
Now, see if you can adapt this example to use Spark Streaming's window functions. For
example, you can compute similar statistics per user over the past minute, sliding every 30
seconds.