Make These Fast-Start Analytics Templates Your New Best Friend

You’re ramping up digital operations. You’ve hired some analytics pros – data scientists, data engineers. You’ve done due diligence on machine learning tools, models and compatible databases. Time to start building your analytic apps.

From a 30,000-foot view, you’ve got three choices on how to proceed. You can build your own apps from the ground-up, using the various toolkits available. At the other extreme, you can buy pre-built apps that come close to fitting your needs. Or you can take a middle-ground approach by using progressive, fast start analytics templates.

There’s no one-size-fits-all solution for every business. But you owe it to yourself to investigate a progressive-template strategy, particularly if your organization is facing:

  • Regular operations and technology upgrades;
  • Uncertainty about future analytics needs;
  • Evolving makeups of analytics teams; and
  • Time-to-repair or time-to-market demands by customers.

Build Or Buy, And Little Room For Error

The build-or-buy question has dogged application development for decades. But with the advent of analytics-everywhere it’s become more critical than ever.

Analytics apps need to look at terabytes and petabytes of data; they need to create massive data models; and they need to do so at speeds that approach real-time. There’s little room for error, and even less room for making a wrong choice in an analytics framework.

Building analytic apps from scratch can waste time and energy – and developer costs – because of the fast-moving nature of change. And change is a fact of life as business transforms to all-digital operations. It can be near impossible for in-house programmers to chase today’s fast-moving marketing targets using conventional development techniques.

At the other extreme, pre-built analytics apps do a good job of saving up-front programming costs. But once installed, they are susceptible to change, too, since they may require expensive and time-consuming vendor upgrades to keep up.

How Vitria’s Templates Work

Vitria offers a middle ground to app-building with a set of five, progressively linked templates. Each template includes source and target interfaces, data prep functions, database schemas and ML models.

Each template fits a specific analytics need, from real-time operational visibility and anomaly detection to change management, incident life-cycle automation and dynamic failure prediction. Also, each template builds on the foundation set by previous templates to help facilitate deployment.

The templates supply 70-80 percent of the functions necessary for their specific applications, leaving the rest to be inserted by using Vitria’s component-based development tools.

The templates’ extensibility means they can be used by different companies, a telecom, manufacturer, or system integrator, for instance, and quickly modified to their – or their customers’ – needs.

Here’s a closer look.

  • Real-Time Operational Visibility: Supports time-series real-time and foundational data curation and schema of digital operations.
  • Advanced Anomaly Detection: Supports methods to detect anomalies within simple and complex digital operations time-series.
  • Change Management: Supports complete monitoring change activities and their impacts on populations across digital operations (e.g. OEE, Support Requests, etc.)
  • Incident Life-Cycle Automation: Supports the correlation of anomalies into incidents in order to reduce noise, better prioritize work and reduce incident life-cycle time.
  • Dynamic Failure Prediction: Supports predictions of potential failures based on correlated anomaly patterns (e.g. set of alarms) and proceeds to proactive maintenance.

The best part: you can get off to a fast start with operations-wide visibility, then build from there as your needs and sophistication grow.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>