We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.


Democratizing Analytics the Easy Way: A Q&A John Thuma of Teradata Aster

Originally published June 24, 2015

This BeyeNETWORK article features Ron Powellís interview with John Thuma, Director of Teradata Aster.† John and Ron discuss Teradata Aster and the very useful AppCenter.
John, Whatís your role at Aster?

John Thuma: Iím a practitioner. Before we called this field big data, it was called very large databases, and I have about 30 years of experience working with very big data implementations. And now big data is a hot field. Iím really into Aster strategy and adoption. I used to run the most advanced solution architecture proof-of-concept projects across the whole spectrum of different use cases.

Aster is a rapid analytics development package. Itís really a platform for rapid analytics development across many genres of analytics such as graph, machine learning, text, path and pattern, and statistical analysis. You can do a support vector machine. You can do it in NaÔve Bayes against petabytes of data within Aster rapidly. Then you have this other facility called AppCenter. AppCenter is a big data consumption service. Itís a website in the Aster environment Ė just a web form, point and click. It allows business people to consume big data analytics and these different types of offerings.

Where do they go to do this?

John Thuma: They log into a web page that is on the Aster appliance, and then just point and click. Itís as easy and pervasive as using your iPhone Ė a very easy experience.

So are you a data scientist?

John Thuma: I am a data scientist. I have been involved in many different types of projects across retail, banking, manufacturing sensors, voter/politics and insurance Ė itís been a crazy wild ride and a lot of fun.

Many people are looking at operationalizing analytics. Where would they start?

John Thuma: This is the most important topic. Just because you can do analytics, doesnít mean theyíre useful. Making an analytic useful means that you can integrate that answer. You can integrate that churn list. You can integrate that next-best offer list into some other system. Itís really about the iron triangle.

What is the iron triangle?

John Thuma: Excellent question. Itís people, process and technology Ė everything weíve learned in the past 30-40 years of computing is still applicable today. Aster is a rapid, easy-to-use infrastructure. It allows you to focus on what really matters Ė changing business and business processes. If you develop a churn score and you want to integrate that into a call-center application or business process so you can actually start to save those customers, you have to change the business process. You have to change people. People are skeptical about these new big data assets and the answers that are coming out. They donít understand them. Even if the prediction algorithms were 100% accurate, humanity Ė just by the nature of its skepticism Ė would doubt them 30% of the time. There is a lot of time spent with people and processes. The reason Aster is so valuable is because itís an easier technology. It allows you to lower your complexity and lower your code surface area. I have something called SQL MapReduce. I have something called SQL Graph. I have all these different genres of analytics, and it follows an ANSI SQL paradigm to implement them. I can take someone thatís a pretty good ANSI SQL developer, and I can turn them into a machine-learning guru in about six months and really make them powerful. And, because they already know how to use data-definition language and they know their way around databases, theyíre going to be right at home with Aster. The reason that is so critical is that it reduces the friction to getting to a solution. And when I can reduce my friction to gain a solution, I can spend less time developing the solution. That means I can spend more time with the people and the process.

One of the big contentions for companies that are implementing analytics is the skill level. So many companies are looking at that and saying they canít find the people. It almost makes it sound like Aster takes away that need.

John Thuma:
It does. The Wright brothers built and designed the first airplane, and they also flew it. Thatís the same way it used to be with these advanced analytics. The person who wrote NaÔve Bayes Ė it was a massive algorithm probably written in C++ or Java Ė was also the person that was smart enough to be able to implement it. Now that person is usually a PhD out of Stanford or MIT or Cornell. Theyíre extremely rare and extremely valuable people. Well, we can take their good work, put it into Aster and make it something thatís democratized and consumable by a whole different set of developers. Also, through our AppCenter, we can now open those analytics to the actual business users as long as they understand what they mean. You still have to understand NaÔve Bayes. You still have to know what support vectors machines are and everything else that goes into machine learning. But the point is that it democratizes it.

So that is from the development side. But when you talk about the AppCenter, what really excited me was it took several of the applications that are consistent across most organizations and created an app so the business users themselves could take that app and move it forward. You still need somebody with some technical expertise.

John Thuma: Absolutely. The data scientist would build the app. Make it reusable. Perfect it. They could actually produce an application that a business user could use. And itís very simple to do. You basically fill out the form and hit ďrun.Ē Then it runs, and they have a variety of visualizations and all kinds of different ways of looking at the outputs. We have a sankey flow visualization. We have a Sigma chart, which is a graph. We have hierarchical diagrams. You can even produce a† raw table if you want. All the other traditional bar charts and things like that are at your disposal as well.

The most important thing is that big data is not only an opportunity for visualizations and consumption of these things, but also an integration. By being able to move that data into that CRM system or that call-center application so I can take the churn score and when that customer calls in, I can see that the consumer is a 90% churn risk. And then what do we do with them? We have a next-best action, and we say letís get them into the customer retention process so we can start to retain them. And that is operationalization of an analytic. There is a lot more to it. Because if youíre not doing that, you have a science project. And science projects are cool for short periods of time. But when you have something that could add revenue to a companyís bottom line, thatís what matters. Thatís what weíre looking for with operationalization. Itís also monetization and the ability to turn these analytics into something that has value to an organization.

So, John, what is the typical time involved for a company that brings in Aster and wants to get one of these AppCenter apps up and running?

John Thuma: Itís highly dependent upon a lot of data issues Ė the esoteric nature of data and quality and complexity and harmonization. But, I have customers that Iíve worked with that took a 12-month development cycle to get an analytic done down to 30 days. I hadnít talked to that customer for a while and I called to ask how things were going. He said he was worried about Aster. I thought there might be a problem. He said, ďItís working too well. We actually have to staff up. We need more people.† We were able to get our projects done in about a twelfth of the time we would have spent with the other, traditional platform.Ē With Aster, itís not about the runtimes. Our runtimes keep up with anybodyís. Itís about the speed at which you can get things done. For instance, just a sessionization with Aster goes from 4,000 lines of code in Java thatís highly stove-piped to your data, not reusable, to Asterís sessionized command, which is one statement in SQL. Select star from sessionized. Itís extremely easy to use. So when Iím coding less and reducing my code surface area, I get to focus on my answers Ė focusing on the ďwhatĒ not ďhow.Ē I get to focus on getting things done and not writing Java code or not moving my data to the analytic because the analytic and the data are together.

Itís a really interesting time. Aster is the best kept secret in Teradata. We have to get it out there into the world. I would love to see people come to our community that we recently launched. Anyone is free to come in and download a free version of Aster Express. You can take the tutorials. I have about 7 hours of how-to videos where I actually teach you in 10 minutes how to do a support vector machine. We actually go through the code, how support vector machines work, and we take you through it all. The URL is Aster-Community.teradata.com.Weíd love people to come out and you can even request a demo on it.

Youíre sure making this easy, John. Thank you for explaining Teradata Aster and the AppCenter.

  • Ron PowellRon Powell
    Ron is an independent analyst, consultant and editorial expert with extensive knowledge and experience in business intelligence, big data, analytics and data warehousing. Currently president of Powell Interactive Media, which specializes in consulting and podcast services, he is also Executive Producer of The World Transformed Fast Forward series. In 2004, Ron founded the BeyeNETWORK, which was acquired by Tech Target in 2010.† Prior to the founding of the BeyeNETWORK, Ron was cofounder, publisher and editorial director of DM Review (now Information Management). He maintains an expert channel and blog on the BeyeNETWORK and may be contacted by email at†rpowell@powellinteractivemedia.com.

    More articles and Ron's blog can be found in his BeyeNETWORK expert channel.

Recent articles by Ron Powell



 

Comments

Want to post a comment? Login or become a member today!

Be the first to comment!