====SampleStream Example====

In this case, we're going to perform a search for the last time Turbine #1 from the Optrix **Wind Farm Demo** was offline for more than a second.

First, we import the libraries we need...

<code python>
import ardiapi
import samplestream
</code>

Then we connect to our ARDI server...

<code python>
srv = ardiapi.Server("demo.optrix.com.au/s/ex")
</code>

Now we create an AQL query to get our data.

<code python>
query = "'Wind Turbine #1' ASSET 'Power' PROPERTY VALUES"
</code>

Next, set up our [[class_samplestream|SampleStream]]. In this case we're going to look at our data one minute at-a-time, with one-second resolution.

<code python>
sstream = samplestream.samplestream(srv,query,60,60,-60)
</code>

Next, set up the //span// of our search. We're going to want to look back up to four days.

<code python>
sstream.Span(datetime.datetime.now() - datetime.timedelta(days=4),datetime.datetime.now())
</code>

And finally, we need to kick it all off!

<code python>
stoptime = None

#Loop until we find an outage or hit the end of the span
while True:
   #Get the next sample
   sample = sstream.NextSample()
   if sample is None:
      #We reached the end of the span
      break

   #If power dropped near-zero over this 30 seconds, we've found our target
   if sample.data['Wind Turbine #1 Power'].min() < 0.1:
      stoptime = sample.time
      break

if stoptime == False:
   print("No Stoppage Found")
else:
   print("Stopped At " + str(stoptime))
</code>

This code walks through the samples until it finds one with a minimum value of 0.1. The power dropping to zero tells us the turbine was offline.

===Other Examples===

This is an example of a problem with //unbounded timeframes// - where we wanted to find something without having a good idea of when it happened. Because SampleStream queries the range piece-by-piece and only if needed, you can scan large time-ranges for events.

Another key application is with //very fine time resolution//, where you want to scan a particular period of time for events that are measured in milliseconds. Even though you'll usually be working with fixed periods of time, the volume of data is still very large - SampleStream allows you to be efficient with memory and server resources, and drop-out early if you've found what you're interested in.

And the final application is in AI - if you want to predict, model or classify something using a Machine Learning model that uses data over time, SampleStream makes it quite easy to extract the data to train and test your model.

See the [[samplestream:view_the_documentation|SampleStream Class Reference]].