Opportunities with the internet of things abound, and generally speaking those opportunities make themselves available to those taking action.
In that spirit, we share five IoT resolutions for 2017, a starting point to get your organization on track for IoT success.
1. Capture a new data source
Every day, companies capture data from interacting with customers and suppliers, as well as third-party data based on the economy, weather, social media and more. Here is how to get going:
Find an entirely new data stream
Set a plan to capture a new data source for your organization. For example, some industrial equipment may already have the ability to output information but it might not be captured today. Or readily available public data might be easily integrated and correlated with current information.
Add structure to an existing stream
You may have an existing source of unstructured data that is not particularly useful in its current form. Taking that same data and adding enough structure to make it accessible to others in the organization can bring new insights.
2. Scope a new application
Brainstorm a new application that delivers new customer benefit or operational efficiency. Options include:
- A new mobile application
- A user experience boost by delivering more accurate and relevant information
- Time-saving tools for customers and the internal business
While planning and building, ask yourself:
- What combination of data sources will provide the most value?
- Can this application benefit from real-time data?
- Can I move to a push model instead of just a pull model for application interactivity?
3. Build an IoT analytics application
Analytics on a fresh view of existing or new data helps drive a business forward. Consider applying existing machine learning models to existing workflows, or applying models to new incoming streams of IoT generated data.
For example, many machine learning models, or in earlier parlance statistical models, can be exported using the Predictive Model Markup Language, or PMML.
Specifically, tools like SAS export models to PMML that can be integrated directly into real-time pipelines. Modern transformation tiers like Apache Spark and distributed databases like MemSQL can natively host these models so that incoming data can be scored in real time.
Architects can expand on the popularity of libraries such as MLlib and TensorFlow to create predictive analytics applications using these tools.
4. Ensure the right foundational data infrastructure
Successful IoT deployments need to span from edge data collection all the way to the data center. Companies like OSI Software provide just one example of collection tools to help feed data into your pipelines.
Once in the data center, a common architecture involves integrating the following tiers.
At the messaging layer Apache Kafka and AWS Kinesis are popular options to aggregate data streams, connecting producers and consumers of information.
Most data pipelines require modifying the data from its state at capture to its state for long term persistence. Converting sharding schemas so data is properly categorized can take place at the transformation tier.
The most accurate model for predictive analytics involves both real-time and historical data, so being able to persistently retain data, including records over time, sets the proper context.
Nothing says “wow” like a real-time dashboard that enables quick visualizations of current data. Popular business intelligence dashboards like Tableau, Zoomdata or Looker, along with custom dashboard options using frameworks like D3.js, allow companies to provide widespread access to fresh data.
5. Set an organizational model for IoT success
There is no question data plays a more important role in today’s business climate with everyone clamoring to “transform.” New CxO roles like the chief data officer and chief analytics officer make that more apparent than ever.
At the end of last year, Gartner estimated that 25% of large global organizations had already hired a chief data officer. By 2019, Gartner expects that number to reach 90%.
Further, Gartner sees a rise in advanced analytics:
By 2018, Gartner predicts that over half of large organizations will compete using advanced analytics and proprietary algorithms, disrupting entire industries. This, in turn, is being driven by the proliferation of devices, connected “things,” connectivity and computing power — all of which creates more opportunities to collect data, analyze it, and potentially monetize it.
There is no better time like the present to get started on your IoT infrastructure planning.
Happy IoT Year!
All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.