How-To: Simulate Time Series Data with Python

j12y created · Feb 28, 2018 at 11:20 AM

PROBLEM

You want to populate your Predix Time Series instance with some sample / test data. If you don't already have sample data in a file you may need to generate some random data.

Python can be helpful for this sort of quick and dirty task.

STEPS

(1) Install the Predix Python SDK

If you don't already have PredixPy, install it with pip. More instructions can be found from the PredixPy Github Repository.

(2) Create a UAA and Time Series Instance

Perhaps you already have instances, but if not the Predix Python SDK can help create them.

 import predix.admin.app
 admin = predix.admin.app.Manifest()
 admin.create_uaa(uaa_secret)
 admin.create_client(client_id, client_secret)
 admin.create_timeseries()

(3) Generate a random value

There are many ways to generate a random number in Python and make it fit within the parameters you are looking for. If you'd like to add more examples that are useful I'd like to see them. Once you have your random value, then you can simply send it to Time Series in the way that makes sense for your application testing.

 import time
 import random
 import predix.app
 
 min_voltage = 200
 max_voltage = 240
 hertz = 1.0/1.0
 
 app = predix.app.Manifest()
 ts = app.get_timeseries()
 while True:
       try:
 
         # Generate a random voltage value simulated from multiple sensors
         # and queue them up.
         for tag in ['Voltage:Sensor1', 'Voltage:Sensor2', 'Voltage:Sensor3']:
             value = random.randint(min_voltage, max_voltage)
             ts.queue(tag, value)
             print(tag, value)
 
         # Send readings from all 3 sensors to Time Series and then wait to
         # take another reading.
         ts.send()
         time.sleep(hertz)
 
     except KeyboardInterrupt:
 
         # Provide evidence data was being stored before exiting
         print(ts.get_tags())
         break