Three steps to analytics success
Analytics that works for your business (not the other way around)
Organizations are increasingly adopting predictive analytics more broadly. However, many analytic teams rely on approaches and tools that won't scale to the level they require. These organizations need a repeatable, effective, efficient process for creating and deploying predictive analytic models. In other words, they must operationalize analytics. The most powerful examples of success use decision management to deploy analytic insight into day-to-day operations.
STEP 1: Understand and solve the right problem
A key challenge is ensuring that the team understands the business problem. Some questions to think about while developing your model plan:
- How will I measure success?
- What data is applicable?
- Who will use the model, and what decisions will it influence?
- Where will I need to deploy this model?
An effective analytics team works in close collaboration with their business partners to answer these key questions and stays in sync through the model development and validation process.
Understanding the problem with decision management
The decision management approach creates a shared framework and collaborative environment for the business, IT and analytics teams. Within this framework, they can better understand the problem by identifying and prioritizing operational decisions.
Decision management links those decisions to the business drivers and performance measures that have the most impact on the business. Modeling the decisions to be analytically improved clarifies and focuses the problem statements to ensure a strong and lasting ROI, and effectively applies constrained analytic resources.
STEP 2: Communication and collaboration
This business-oriented approach creates productive interaction between the analytics and business team and clarifies how and where the resulting analytics will be deployed. This helps guarantee a successful deployment once the models are developed.
Today, the results of the predictive model development process are often captured in a document, report or dashboard, or in code generated by the modeling system. For strategic decisions, a report is often sufficient since it provides the insight to review the analysis and make a decision. But to harness the full value of predictive analytics, analytic insights need to be deployed in operational systems.
Taking the next step
With industrial-scale model development processes, decision management enables a reliable architecture for deploying the models into operational systems. Making faster and better decisions requires the integration of decision logic and analytics in operational systems.
Decision management ensures that when a model is built and validated, all the teams know where and how the model will be deployed. The teams can now focus on the point of operational decisions and the systems needed to make them.
Using an in-database infrastructure
One way to make predictive analytic models available in operations is to use in-database scoring infrastructure. These push analytic models directly into the core of your operational data stores. Once deployed, the models are available as a function or SQL and can be included in views or stored procedures. Operational systems now have access to the model's results while ensuring those results are based on current data, and results are calculated live and when requested. Business rules can then access the predictive analytic result like any other piece of data.
Direct analytic deployment
Predictive analytic models can also be deployed directly into operational environments using decision services: implementation of a decision and how other systems will find out what the decision is. A decision service makes the decision reusable and widely available. Decision services are essentially business services in a service oriented architecture (SOA) that deliver an answer to a specific question. Because they don't make any permanent changes, they can be used to answer questions whenever they arise without worrying about potential side effects.
Regardless of how predictive analytics is deployed into the decision service, it should be used in real time to ensure the use of the most current data. Analytic teams must recognize that modern operational systems are real-time.
STEP 3: Reliable deployment
Decision management focuses on decision effectiveness, and more effective decisions reduce fraud, improve customer interactions and spot opportunities. This step focuses on implementation processes of a reliable deployment architecture. An industrialized process incorporates five key characteristics:
Systematic approach to data management
Organizations need to be able to use transactional data and event streams, unstructured or semi-structured data and more to understand what customers are doing. A systematic approach to information management encompassing data quality, integration and management ensures standardized, efficient access.
Predictive analytics workbench environment
A more systematic approach feeds into a workbench environment for defining the modeling flow, thus streamlining and standardizing how analytic modeling is performed. In-database mining capabilities integrated with the workflows can push data preparation, transformation and modeling algorithms into the data infrastructure, improving throughput by reducing data movement. In-memory and other high-performance analytic capabilities and intelligent automation can be applied as appropriate.
Engagement of less technical users
Features that allow less technical users to build and execute workflows take advantage of underlying automation capabilities to produce large numbers of "good-enough" models quickly. Working collaboratively with an analytic team, these users can produce first-cut and less core analytic models, participate more fully in reviews of models and allow the analytic team to focus on high-value, high-complexity problems.
Fast model turnaround
A high-performance analytic infrastructure scales to deliver high availability, workload management and scheduling. Products that create a distributed grid environment provide parallel job execution across multiple servers with shared physical storage. Algorithms written to take advantage of this infrastructure make dramatic performance improvements possible.
Management and monitoring of models
Certain capabilities allow an analytic team to set up automated monitoring of models to see when they need to be retuned or rebuilt. These capabilities also help track the performance of models to confirm their predictive power and behavior. This workflow too can be defined and managed, bringing even those not using the predictive analytic workbench, such as database administrators and integration specialists, into the process.
Today, many organizations have ad hoc processes for business problem definition. It can take months for predictive models to be deployed; some are never deployed at all; and others are deployed but not monitored. This can result in models that degrade to the point where they do more harm than good.
Predictive analytics can deliver insight and predictions, and resolve business uncertainty into profitable probabilities. Decision management combined with a robust and modern technology platform ensures that analytics is solving the right problem; delivers a repeatable, industrial-scale process for developing models; and provides reliable architecture for deploying predictive analytic models into production systems.
The GE C&I Consumer Home Services Division estimates it saved $5.1 million in the first year of using analytics to detect suspect claims. By operationalizing analytics that predicts the likelihood of fraud, GE now automatically audits 100 percent of the claims for suspicious activity.
The predictive analytic models embedded in the operational environment evaluate facts about the services performed including the product, customer, complaint or work completed to predict the likelihood that a claim is fraudulent. The audit team can then focus their efforts on the highest-risk claims.