Speaking at last year's Big Data & Analytics for Pharma Summit, Vijay Nandakumar, Chief Executive Officer of D Cube Analytics, shared the benefits of leveraging a more productized approach to big data in order to provide analytic solutions.
Before you can begin to attempt anything as ambitious, the first step is to identify the common problems areas data analytics professionals face daily. "As part of incepting D Cube, we talked to a number of folks who are the common stakeholders in any data analytics implementation," Nandakumar starts. "You know, those would be your enterprise IT folks, your functional vendors like Salesforce Adobe or other consultants and, the end business users themselves.
"What we found, and I know I'm preaching to the choir here, is a list of challenges that crystallized out of that survey. There was a distinct lack of domain context in terms of implementing data analytic solutions."
The issue is that data analysis is a wildly large bracket and the struggles being faced can vary wildly. "There are a lot of capability silos in place in terms of multiple vendors coming in with multiple capabilities. For example, one vendor comes in with data management capabilities, another with visualization, advanced analytics etc. It's oftentimes very hard to implement them in an integrated format."
And once you take into consideration all the issues that come along with the scalability of newer data sources from a volume standpoint, a set of challenges become evident. "Organizations being saddled with expensive contracts that are driven through the legacy model of analytics implementation. You have solutions that are misaligned with the original business objectives. Also, the improper orchestration of the solutions wherein one capability often is not in sync with the other capabilities. And finally, a heavy dependence on resources which again accrues cost over the long-term."
In order to combat this, Nandakumar and his team developed the Digital Discovery Suite platform. "This basically leverages and productizes a lot of the domain data knowledge and business processes into a bunch of axial raters that disrupt the cost-time framework of a typical analytics implementation," he explains.
To pull this off, the platform was driven by a few "foundational components":
1. A strong data management foundation: This "takes care of all the heavy lifting required to get data from multiple sources, wrangling them and getting them into an analytics-ready format.:
2. Pre-built extractors: These drew from "industry standard syndicated data sources, even unstructured data sources like web and survey data. All of those things are already pre-built and getting into a format where it's ready to be analyzed."
3. Modular analytics workbench: "This is nothing but a series of out-of-the-box dashboards, KPIs and, models that basically power the different use cases across the pharmaceutical value chain."
4. The data science sandbox: This completes the platform and is a "highly visual and intuitive data advanced analytics platform. It is integrated into our foundational data engine and this can basically take care of all advanced analytics needs right from basic regressions, to your advanced predictive models."
As the D Cube team was fundamentally focused on the commercial side, they had a number of use cases built out right from "a pre-launch phase to the entire lifecycle management phase". Nandakumar continues, "a lot of the clinical development use cases are currently in development. We also have a self-service analytics workbench as part of the platform and this is something that enables a lot of the power users within the organization to drive their own ad-hoc analytics or do their own data wrangling to set up the solutions they want."
The benefits of this more integrated approach have been impressive. "We have been able to deploy projects for our customers up to 50% faster than the traditional mode of deployment," Nandakumar says. "We are also able to save quite a bit of cost, sometimes up to 40%, over the total cost of ownership over a year or two time-period.
"Also, the way the solution is architected is, it comes with inbuilt seamless scalability from the handling data volume and variety standpoint. Our platform is infinitely customizable in the sense you can basically come up with any use case and we can basically power that through the platform."