dawidwys commented on a change in pull request #7177: [FLINK-7599] [table]
Support for aggregates in MATCH_RECOGNIZE
URL: https://github.com/apache/flink/pull/7177#discussion_r237569734
##########
File path: docs/dev/table/streaming/match_recognize.md
##########
@@ -211,6 +211,66 @@ If a condition is not defined for a pattern variable, a
default condition will b
For a more detailed explanation about expressions that can be used in those
clauses, please have a look at the [event stream
navigation](#pattern-navigation) section.
+### Scalar & Aggregate functions
+
+One can use scalar and aggregate functions in those clauses, both
[built-in]({{ site.baseurl }}/dev/table/sql.html#built-in-functions) as well as
provide [user defined]({{ site.baseurl }}/dev/table/udfs.html) functions.
+
+Aggregate functions are applied to subset of rows mapped to a match. To
understand how those subsets are evaluated have a look at the [event stream
navigation](#pattern-navigation) section.
+
+With a task to find the longest period of time for which the average price of
a ticker did not go below certain threshold, one can see how expressible
`MATCH_RECOGNIZE` can become with aggregations.
+This task can be performed with the following query:
+
+{% highlight sql %}
+SELECT *
+FROM Ticker
+MATCH_RECOGNIZE (
+ PARTITION BY symbol
+ ORDER BY rowtime
+ MEASURES
+ FIRST(A.rowtime) AS start_tstamp,
+ LAST(A.rowtime) AS end_tstamp,
+ AVG(A.price) AS avgPrice
+ ONE ROW PER MATCH
+ AFTER MATCH SKIP TO FIRST B
+ PATTERN (A+ B)
+ DEFINE
+ A AS AVG(A.price) < 15
+ ) MR;
+{% endhighlight %}
+
+Given this query and following input values:
+
+{% highlight text %}
+symbol rowtime price tax
+====== ==================== ======= =======
+'ACME' '01-Apr-11 10:00:00' 12 1
+'ACME' '01-Apr-11 10:00:01' 17 2
+'ACME' '01-Apr-11 10:00:02' 13 1
+'ACME' '01-Apr-11 10:00:03' 16 3
+'ACME' '01-Apr-11 10:00:04' 25 2
+'ACME' '01-Apr-11 10:00:05' 2 1
+'ACME' '01-Apr-11 10:00:06' 4 1
+'ACME' '01-Apr-11 10:00:07' 10 2
+'ACME' '01-Apr-11 10:00:08' 15 2
+'ACME' '01-Apr-11 10:00:09' 25 2
+'ACME' '01-Apr-11 10:00:10' 30 1
+{% endhighlight %}
+
+The query will accumulate events as part of `A` pattern variable as long as
the average price of them does not exceed 15. Which will happen at `01-Apr-11
10:00:04`. The next such period that starts then will
+exceed average price of 15 at `01-Apr-11 10:00:10`. Thus the results for said
query will be:
+
+{% highlight text %}
+ symbol start_tstamp end_tstamp avgPrice
+========= ================== ================== ============
+ACME 01-APR-11 10:00:00 01-APR-11 10:00:03 14.5
+ACME 01-APR-11 10:00:04 01-APR-11 10:00:09 13.5
+{% endhighlight %}
+
+An important thing to have in mind is how aggregates behave in situation when
no rows where mapped to certain pattern variable. Every aggregate, beside
`COUNT` will produce `null` in those cases. `COUNT` on the other hand will
Review comment:
You are right, didn't think of UDAGs. Do you think it is worth adding such
disclaimer at all. I've added it in the first place, because there was
something similar in the SQL standard, but there are no UDAGs there.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
