UV Computation

[qingyuanxingsi]

Given a log(dataframe) containing the purchase records of users, how can i compute the unique count of goods purchased by a certain user?(Supposing that one can purchase a certain good multiple times) Would you be so kind to give me some demo? Much thanks!

[tovbinm]

It's a classical case for aggregation. Simply aggregate the dataframe by (user + item), e.g. https://mungingdata.com/apache-spark/aggregations/

[qingyuanxingsi]

@tovbinm Oh,I was wondering if there was any build-in support? I want to use with the window operator?I found sum aggregator in the code, but failed to find the count(distinct) operator?