Recap WITI Orange County - September 18, 2014

The Secret Life of a Data Scientist - Real Life Analytics and Data Science
Sept 18, 2014

With over 80 people in attendance at the Eureka Building in Irvine, CA. Helen Norris, Vice President and CIO, Chapman University (and a former WITI Network Director for Sacramento) kicked off the evening as the moderator, describing the role of a data scientist as the sexiest job of the 21st century! Helen introduced the panelists and their relationship with ‘data':

Dr. Erik Linstead, Assistant Professor of Computer Science in Chapman University's School of Computational Science

Parm Lalli, Director, Sunera, leads the multinational data analytics practice

Barbara Forth, Principal Information Architect, Capital Group

Dr. Linstead opened the discussion on how the role of a data scientist can be a very interdisciplinary role as he works with Biologists, Chemists, Mathematicians, and other disciplines. Dr. Linstead and some of his Ph.D. students are working on a potential healthcare solution. Their program is called AMP - the “Autism Management Platform.” Functioning in some ways similar to an online news feed, the system allows for personalized, private, specifically-directed information sharing between caregivers. One in 68 children is diagnosed with autism each year. AMP was developed to provide communication via cell phone and an Internet portal to help track and monitor behaviors, day-to-day routines, schedules, milestone achievements and activity patterns through caregiver input into the system.

Parm Lalli has been in the data analytics business for 10+ years and is an ACL Certified Data Analyst and a Certified Information Systems Auditor. He has led several Continuous Controls Monitoring implementations with many data analytics opportunities. Parm has also worked with ISACA (Information Systems Audit and Controls Association (www.isaca.org) to determine how to use data for fraud detection. Parm used analytics to assess compliance with the FCPA (Foreign Corrupt Practices Act) which addresses accounting transparency requirements under the Securities Exchange Act of 1934 as well as concerning bribery of foreign officials. He developed analytics from publicly available data from the CIA to track what government officials were doing. He reminded us that companies need to be aware of how people may use publically available information.

Barbara Forth has 25 years of data management experience with Capital Group and Fidelity Investments. Barbara is championing an Enterprise Information Management program and is focused on establishing a company-wide Data Governance Practice.

Helen, our moderator, formerly with the California State University system, leveraged data and analytics to determine the cause for students having difficulties in graduating. Use of analytics could potentially help with intervention to get students back on track to graduate.



What does Big Data mean to you? Variety, Volume, Velocity, Variability - in essence, it is just massive amounts of information that's difficult to understand with current tools. Barbara offered that it forces firms to rethink how they do application development. IT shops can shorten development lifecycle by providing the means to the data, then letting data scientists come up with the requirements to develop the reports.

Formerly, firms always focused on cleansing the data first before reporting. Today, reporting can be done by just bounding the error. One can use crude modeling techniques and leverage data with some boundary, for example, 75% accuracy.

Big data is used today, perhaps without you knowing it. Stores can monitor placement of products in store aisles based on the location of your iPhone and how long you remain in an aisle. While Facebook and Gmail are ‘free' for your consumption, YOU are the product for them. Ads are now being centered around YOU. In the very near future, if you are shopping online and look at an item several times, retailers may ship the product to the warehouse assuming your intent to buy.

Available tools: Lots of open source tools and high performance computing platforms are now available. Tools such as Hadoop, Octave, Revolution R (statistical programming language) allow big data to be comprehensible. Parm mentioned that he may have previously run 10k records per second, he now needs 20M records per second. Traditional systems wouldn't allow for this type of speed in the past. The new platforms like Hadoop or NoSQL (and several of the new tools), allow you to distribute processing and reporting of millions of records across several servers to deliver performance for analysis.

What changes do Organizations need to make to address this new technology and business opportunity? A mindset shift is required - it's extremely important to think about the business process, not just the data. Business architecture functions are highly unfunded which makes it difficult to think about business capabilities before trying to solve for data required. Furthermore, companies need to have a corporate governance structure. It's important to know the data you own and what you don't own. Most firms think IT owns the data. This is typically not the case. In the medical fields, data ownership is reversed. Clinicians think they own the data, but it should be the patients. Often, the practitioners don't have your medical file. Why not have the patient bring their medical file?

Firms need to get a handle on upstream data and how regulations (i.e.HIPAA), master data, and dark data (e.g. data we don't understand) factor into data governance. When sending data to 3rd parties, firms need to be able to send the data and ensure the provider cannot infer anything in an aggregated form. When sent individually, the data may not be considered sensitive. Data sharing agreements may help. There's also the concern around scrubbing data too much that it may lose meaning. One example Dr. Linstead mentioned was the removal of zip codes which impacted the ability to analyze autism cases with patient proximity to oil refineries. Statistical cleansing frameworks can help make sure you don't lose too much meaning.

Firms need to get to the point of doing predictive analytics through forecasting. What will the data be tomorrow? Firms have problem today with not being able to tie external data with their internal data.

What skills are required to be a Data Scientist? Mathematics, Statistics, Logical thinking, Knowledge in a specific business area, and Communication skills. Strong communication skills are critical in order to understand the business and translate IT to the business. The tools and buzz words change over time, but the foundational skills in the math arena are required. Having a fearlessness with math and statistics is a must! The day job of a data scientist is to apply models to data and condense/categorize it - i.e. take the noise out. It's important to keep in mind that business problems need to be solved by teams, not just a single person.

The evening wrapped up with a raffle with several give always, and a reminder for the next event October 16.

Special shout outs to Georgia Zachary, Kimberly Hatten, and the Women of AT&T for sponsoring the event and to Chapman University and Dr. Michael Fahy, Dean of the Computer Science Dept. We appreciate Dr. Fahy and Dr. Linstead support, bringing their students to interact with working professionals.

Opinions expressed by the author are not necessarily those of WITI.