KDnuggets Home » News » 2015 » Feb » Opinions, Interviews, Reports » Interview: Rachel Hawley, SAS on the Quest for Agile Analytics ( 15:n04 )

Interview: Rachel Hawley, SAS on the Quest for Agile Analytics


We discuss Agile Analytics, moving from traditional Analytics to Agile, challenges in operationalizing Analytics, SAS Enterprise Decision Management and SAS In-Memory Statistics.



rachel-hawley-sasRachel Hawley is an Analytical Solutions Architect at the SAS Institute specializing in high-performance analytics solutions. Rachel consults with executives and analysts, works across industry verticals, and has seen a variety of challenges and applications throughout her career. Most recently Rachel spent several years working with government agencies to architect solutions to combat fraud, waste and abuse.

Rachel has a passion for teaching, presenting and finding innovative ways to explain difficult and technical topics to non-technical audiences. She first discovered this passion when teaching pre-calculus at NC State as a graduate student where classroom creativity was needed to help students find a new way to learn material that has challenged them in the past.

She holds a Bachelors in Mathematics from the University of Rochester and Masters in Operations Research from North Carolina State University.

Here is my interview with her:

Anmol Rajpurohit: Q1. How do you define "Agile Analytics"? How is it different from traditional Analytics?

Rachel Hawley: Agile analytics allows modelers to have a conversation with their data. agile This means that, unlike traditional analytics where a great deal of time is spent waiting for models to complete running, analysts can get immediate answers to their questions which will then lead to subsequent questions and spur on a back and forth conversation with their data.  Model building and analytics is an iterative process in principle. As the iteration time is reduced, modelers can provide organizations with better answers more quickly.

AR: Q2. What approach would you recommend to cross the chasm between traditional Analytics and Agile Analytics?

RH: There are two great places to start:
  • Empowering more people to be able to leverage analytics with tools where analytic best practices are embedded in a bounded but not black box way.  This means that the entire organization can mature analytically not just a small set of PhD’s or data scientists.
  • Embedding analytics into technology: This means moving the math to the data or the decision point whether scoring inside a database or a Hadoop cluster or putting risk score in the hands of a decision maker.

 
These both enable organizations to make more data driven decisions more quickly.

AR: Q3. What are the biggest challenges in operationalizing Analytics? What questions should organizations ask (or what metrics they should measure and track) to assess the ROI on Analytics?

RH:
From my experience, many customers have a hard time having a single view of the entire analytics process from data access to model building and finally model deployment. 

This is often because the data environment, modeling environment and production environment are all siloed and / or on different platform.  It can require a great deal of manual intervention to move from one environment to the other which can be extremely time consuming.  This results in the analytics getting stale as customer behavior changes and the analytic process can’t adapt quickly enough.
analytical-lifecycle
AR: Q4. What are the key features and capabilities of SAS Enterprise Decision Management solutions?

RH: SAS® Decision Manager takes data, business rules and analytical models and sas-logoturns them into consistent, automated actions that drive faster, better operational deci­sions. And it does so from a centrally managed, easy-to-use interface.  SAS® Decision Manager is used to automate the hundreds and thousands of opera­tional decisions that happen every day. Not only does this make organizations more efficient, but it eases the burden of manually redefining models into production.

AR: Q5. What features of SAS In-Memory Statistics for Hadoop interest you the most? How have SAS clients responded to this product since its launch in the early 2014?

RH: I find the speed at which models can be run as the most intriguing aspect of SAS® In-Memory Statistics. Leveraging SAS’s In-memory technology models can be run at speeds never seen before and this means that our customers can now ask questions of their data they haven’t been able to before.
sas-in-memory-analytics
This product has been very well received since its launch.  Being an interactive programming interface it especially resonates with people who are used to open source technology as that may be what they used in school.  SAS® In-Memory Statistics, however, drastically speeds up the time at which models can be run and thus enables them to have a conversation with their data and, as I like to say, model “at the speed of thought.”

Second part of the interview

Related: