Making simple Esper avg query

Question

I am very new to Esper and I just went through the tutorials, so there are quite a few doubts that I have.

I am trying to make a query that computes the average of the information received by all datasources.

So to start, I have a simple query that just prints out everything I receive:

select * from pattern [every e=MyClass.Event]

Then my next step was to calculate the avg:

select avg(cast(value, float)) from pattern [every e=MyClass.Event]

But Then I get an error:

[ERROR] [2013-12-16 17:12:16,959] [qtp1609813298-11] net.jnd.thesis.helper.RoutesWrapper  - Error starting statement: Property named 'value' is not valid in any stream [select avg(cast(value, float)) from pattern [every e=net.jnd.thesis.camel.bean.CamelInternalEvent(sid=0 and sid<1387214056883)]]

This basically means that while using this format, I cannot access property value. I know for a fact that the stream only contains float numbers, so I tried another version of the query to make the average of everything:

select avg(*) from pattern [every e=MyClass.Event]

Which also does not work because apparently using avg(*) is incorrect syntax.

I have re-read the quick start and tutorials on the website but was unable to find anything able to help me. How do I write a query that makes the AVG, Median, or STDDEV of all the values that I receive from all the datasources?

Answer 1

After you tell esper about the classes you are using in the esper configuration file.

<?xml version="1.0" encoding="UTF-8"?>
<esper-configuration xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
                     xmlns="http://www.espertech.com/schema/esper">

<event-type name="MyClass" class="package.MyClass"/>
</esper-configuration> 

You should be able to easily process a stream of events and compute the average:

select avg(value) from MyClass

Where value is a public attribute of MyClass or has a getter in the form getValue. Note that you will need to get the average data somehow. An easy way is to register an object with an update(float) method when you crete the statement. Esper will call this with a new average each time a new event is received. There is a similar example statement on the Esper solution paterns page that calculates an average for each minutes worth of data. It was not clear in your question why the pattern expression was needed, so I left that out.