Interview: Ksenija Draskovic, Verizon on How to Not Get Lost in the Big Data Wilderness

We discuss recommendations for data-driven decision making, challenges and benefits of using unstructured data, managing expectations and key trends.

ksenija-draskovicKsenija Draskovic is Associate Fellow and Head Data Scientist at Verizon Wireless. She has over 20 years experience in using multichannel data sources to understand customer behavior providing key findings and actionable insights to business stakeholders, embedding those insights into business processes and deploying predictive analytics throughout the organization.

She and her team covers variety of analytics for Marketing, Finance, Real Estate department, including predictive modeling, Big Data integration and unstructured data analytics. Ksenija is a regular speaker at Predictive and Big Data Analytics conferences in US and Internationally.

First part of interview

Here is second part of my interview with her:

Anmol Rajpurohit: Q5. Almost all organizations have Analytics capabilities today. Yet, only a few have been successful in embedding Analytics in all business decisions. What are your recommendations to achieve data-driven decision making throughout an organization?

start-smallKsenija Draskovic: I would start small by doing the first project that addresses the immediate and urgent business need then

- Clearly define the scope of the project and deliverable
- Assemble the team that will work together and champion the project within their own organization units
- Think small but anticipate the few steps ahead, always keep in mind scalability
- Include only data points that are stamped and verified
- As many data points as possible, the more the better
- Communicate the discovery progress and first results
- Quickly show the results and achieved ROI

As a general rule, make it easy for end users to talk to you, welcome questions, simplify the results and provide the recommendation on how the results should be acted upon. Build a trusting and lasting relationship with your end users the same way your company does it with its end customers.

AR: Q6. How did you address the challenge of an increasing amount of unstructured data? What were the key benefits of using unstructured data?

unstructured-dataKD: Unstructured data is messy, loosely formatted, they need to be integrated with structured data to realize their full potential. They can, on their own, uncover a few key pieces but you need to assemble the full puzzle to see the emerging trends and the complete picture.

Big Data although coming in a big volume, velocity and variety rarely cause Big Business changes on its own, they need to be well understood within the overall context to be acted upon appropriately. Their value is in bringing up-to-date time and granularity components that are important for getting high definition insights.

AR: Q7. In Big Data, we often hear stories of over-ambitious targets and poor delivery. How do you manage the expectations of all business managers asking for Data Science solutions?

expectations-realityKD: After the initial start with scalability in mind you should be able to handle the increasing number of requests with no major issues. The key point is to bring all data sources together, to have well established processes that load and reconcile data.

Data scientists shouldn't be burdened with data processing step that needs to be only 20% of the work, the majority of their time should be spent on exploration and research tasks for achieving the best results. Once the unified and reconciled version of the data is available there are rarely any problems in handling new data analytics requests.

AR: Q8. What key trends would you expect to drive the growth of Predictive Analytics for the next 2-3 years?

  • Currently Analytics is established and present in the majority of big organizations, Big Data wave definitely brought awareness and deserves lots of credit for it. I think analytics will penetrate next into medium and small size companies. These businesses may not have capability to do their own in-house analytics but hopefully this functionality will be fully available on the cloud where companies will be able to upload their data and get the results quickly.
  • predictive-analytics-trends
  • What is more exciting is that analytics will soon become personal, we will have our own Data-Science App on smartphone or watch that will advise us things such as, for example, that the fridge is empty and better get some food on the way, also grab 9V battery since the fire alarm in the living room went off this morning. The Internet of Things will further facilitate and expand that personal level of analytics.
  • Also I am expecting that Hadoop, NoSQL, Pig, Python, R, etc. will evolve and become more robust in functionality, user interface and flexibility for advanced analytics on heterogeneous data sets.

Third part of the interview