For this article, Ron Powell, independent analyst and expert with the BeyeNETWORK and the Business Analytics Collaborative, interviewed Roger Fried, senior data scientist of the Aster Strategy and Adoption Team at Teradata. They discussed the importance of IT, business and data science working together to meet the needs of the business.
From a data scientist perspective, the modern organization consists of many silos – IT, business and data science. What challenges do these silos have when trying to work together?
Roger Fried: For data science, it is vital that they work together. Data science is a team sport. It is complicated, and there are a lot of hand-offs. The problem is that many of the key people are in those three silos. IT is where the data is and where data is sourced. Data science is where you process the data, learn it and then add the value. The business has a need, and then whatever the data scientists discover has to be delivered to business to solve those needs. All three of those groups have to work together. If they don’t, then the ball is dropped.
Isn’t it hard enough? We have been trying to get IT and the business to work together. Now we have to bring the data scientists in too?
It’s very much like the old equation where before you would focus on how to get BI
to work. In BI you have the business users, and they talk with IT. They try to restructure the data. Now adding the data scientists and their insights the stakes are higher. The potential value for transforming the enterprise and the ways that the resulting data can help you in your business is tremendous. And, of course, the financial risks if you don’t manage to deliver that value can be high as well.
You mentioned BI. How has your background prepared you for these types of conversations?
Roger Fried: I had a sixteen-year career where I was the intersection between finance and IT, implementing planning and budgeting systems, BI and more. So I was very comfortable with managing that conversation between two of the silos. When I moved into data science, it was a natural progression. On the other hand, it does increase the problems. For me, though, it makes it that much more interesting.
Can you give an example of some of those conversations?
Roger Fried: The simplest conversation for an IT DBA or IT specialist is understanding the limitations of the end users when they’re using a product that requires a database. These include security, access rights, dependency. All of these things are the lifeblood of IT, and they expect the users to understand them. Unfortunately, they don’t. So trying to guide that communication is really important.
How would the conversion work with these different groups or silos concerning the integration of open source R with Teradata Aster for instance?
Roger Fried: I love open source R. It’s a great tool that integrates the larger community – that volunteer, scientist or professor. It unifies all of them in a coherent effort to deal with data. On the other hand, it integrates that with Teradata, which is the best when it comes to enterprise data management. And, it’s fascinating to bring them together, but you have that same basic flow where you have to source your data, you have to pass it through your data scientist, and you have to push it on to the end user. So you have to communicate effectively. You have to explain to the IT person what R is – a tool for the data scientist. You have to explain how it is installed. On the other side, you have to communicate to the data scientist the importance of operationalization. You have to explain that this is how they get paid because this is how the company makes money. You have to guide that communication or that language transition from those three different silos to be able to integrate them and make it work.
What do you enjoy most about your job?
Roger Fried: I just love the fact that I can jump from place to place and deal with what seems to be simple questions and issues that have tremendous value. For example, a couple months ago I was speaking at a conference of state tax officials – IT folks. This is an industry that has used rule-based processes to identify fraud and other issues for years. They were looking for alternative solutions, and the idea of machine-learning – the idea of what you do in the creation of a model and how that can be faster and more accurate and deliver the results they were looking for – has tremendous value for them. It’s these basic ideas where you can see people’s eyes light up as they understand how this makes sense.
If I was a business user, what should I know about transitioning to analytics?
Roger Fried: When I transitioned to analytics, I have to say that I underestimated the journey that I would have to take. I thought I knew big data from my perspective. I thought I knew systems. But you really have to accept, as a first step, that programming is the pathway that you have to walk, regardless of whether there are new-fangled tools and promises out there about easy analytics. Programming is really the test of whether you’re tall enough to ride the ride.
What do you think IT should know about data science?
Roger Fried: First, they really have to understand the types of flexibility that data scientists must have to do their jobs. It’s not a matter of provisioning a set of fields or even a large set of fields for a BI project. IT has to give the individual user the type of access that they’re not accustomed to giving. On the other hand, IT still has to do their job, which is to make sure there is reliability built into the process and that security – even though it’s defined differently – is there. By performing that role, they can rein the data scientists in but give them enough data and flexibility to be able to deliver the value that the business side expects from them.
Do you see most people being able to make this transition from finance to analytics?
Roger Fried: There is a lot of interest right now in finance communities. Finance is trying to understand its role as the organization changes. And, finance is recognizing that, on one hand, there is the threat of automation as they see it. And, on the other hand, there is a tremendous amount of talk about digitalization and analytics. With the types of business questions that are involved with data science, it makes perfect sense that the finance FP&A types – the ones that intersect with the business –should be able to play a role, but it is a much longer journey in many ways than finance tends to understand.
Maybe I look at finance in the wrong way, but I always look at it as being very black and white. And I look at analytics as being a little gray. Is that true?
Roger Fried: As a finance person, I have always divided finance into two buckets. One is accounting. And even if they have finance in their title, if they are the type of person that does very routine processes where they’re gathering up the beans and putting them into the right boxes, then they’re accounting. If they’re the type of person that takes the beans out of the boxes and plays with them to try to make a useful picture, then they’re finance. So certain types of finance roles have a natural connection to analytics. In FP&A – financial planning and analysis – where you’re involved in the trenches of deciding what targets should be and what the budgets should look like, and helping people understand what the financial numbers mean to them – those types of people deal with business questions every day and they’re translating business decisions into money. So they have a very important skill set. They’re the ones who day in and day out are concerned with ROI. If you can’t deliver that ROI from data science or for from your IT project, then you’re not delivering what you need to. So those finance people do have a role, and that financial perspective needs to be remembered.
Great. Roger, thank you for sharing how we can transition from finance to analytics.
SOURCE: The Transition from Finance to Analytics
Recent articles by Ron Powell