May 3, 2017 | Written by: Natasha DSilva
Categorized: Data Analytics | What's New
Share this post:
The IBM Streaming Analytics service is a cloud based service for IBM Streams. Streams is an analytics platform that allows you to create applications that analyze data from a variety of sources in real-time. We announced earlier that the service had been upgraded so that Streams applications could be created in Python. Now, it even easier to develop in Python because of the latest set of upgrades to the service. Also, the Streaming Analytics service now integrates with the IBM Data Science Experience. This means that you can now use models and algorithms written in Python to analyze not just static, but also real time data!
If you use the service, your Streams instance has already been upgraded to the new version and no action is required.
Key enhancements in the Streaming Analytics service
- Developers can create Streams applications in Python without installing Streams: Previously, you had to have a local installation of Streams to create applications that would be executed on the service. Now, you don’t have to install Streams! The Python API has been enhanced so that you can now directly submit your application to the service from your code.
- You can monitor and manage your Streams applications via REST: Check the status of an application and even view data being analyzed by the application using the new Python REST API. You can use the API to check the status of a service instance or cancel a running job.
DSX and Streaming Analytics
“Use models and algorithms written in Python to analyze real-time data without having to store the data first, or install any software.”
As was mentioned before, the Streaming Analytics service now integrates with the Data Science Experience (DSX). DSX is an interactive environment that allows you to quickly extract insights from data. You can create and run your notebooks right in your browser using Python, R, and Scala, Spark, and more.
With the integration of DSX and Streaming Analytics, you can now create Python notebooks to analyze real-time data without having to store the data first or install any software.
Use your models and algorithms on real time data
For example, you could create a Python application to predict the likelihood that a given engine will fail based on its temperature. After creating a model and training the model on historic failure data, you can submit the application to the service to run the model on real time data readings. You could even configure the application to alert you if a failure is imminent.
Retrieve the analysis results from the service and visualize them within the notebook
Create useful visualizations for your streaming data using Bokeh, Matplot, and PixieDust.
For example, below is the output of one of our sample notebooks. It monitors and analyzes patient vital signs and then creates a visualization of the analysis results from the service. As you can see, the graphs are updated in real time with data from the Streaming Analytics service.