TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

Five Minutes with a Chief Architect: Claus Jepsen of Unit4

Yes, we know you probably hate math and statistics. Deal with it -- if you want to be a top notch data scientist, that is.

• By James E. Powell
• October 25, 2016

Claus Jepsen has spent the last few decades developing and architecting software solutions, most recently at Unit4 , where he is the chief architect leading the ERP vendor's focus on enabling the post-modern enterprise. At Unit4, Claus is building cloud-based, super-scalable solutions and bringing innovative technologies such as AI, chatbots, and predictive analytics to ERP.

Claus is a strong believer that having access to vast amounts of data allows us to construct better, non-invasive, and pervasive solutions to improve our experiences, relieve us from tedious chores, and allow us to focus on what we as individuals really love doing.

Upside spoke to Claus about his work as a chief architect.

UPSIDE: Are you working on anything interesting right now? If not, what's your dream project?

Claus Jepsen: Absolutely! One extremely interesting and very inspiring project is our research into creating self-driving software solutions. By self-driving I mean software that can makes decisions and act on behalf of the user based on historic data. Historic data in an enterprise could be workflow actions, like approval/rejection as well as the corresponding transactional document.

We're researching the ability to leverage historic data and by combining the transactional data with workflow actions to determine the most probable action a user would take -- allowing the system to automatically handle the task without end-user intervention.

Another project I'm currently working on is building the next generation user experience: a conversational user interface. We're currently witnessing a paradigm shift in human/computer interaction driven by bots, AI, and predictive analytics. The shift is towards the most natural means of communication, which is vocal and written communication. The technology to decipher spoken and written language is now capable of taking written/spoken direction and turning it into actions, reliving us from dealing with complex, intrusive and (sadly) even poorly user interfaces.

From a machine learning and data science perspective, what's really interesting about this project is using past conversations to anticipate what users are trying to accomplish when interacting with an enterprise solution.

If you could go back in time, what's the one thing you would tell yourself as a new analyst/data scientist?

Well, interesting question. I didn't graduate as a data scientist, as that didn't exist back in 1992 -- so I've learned data science skills on the go. However, as part of that journey, I wish I'd paid more attention when going to statistics classes because the majority of data science (and the field around machine learning) is mostly about statistics! Math is equally important, so knowing your algebra and calculus is essential for understanding how machine learning works.

What's a personality trait you think people need to succeed at your job?

The ability to speak with business owners and/or stakeholders to really understand what they're looking for. The problem with data science and big data is that it's not worth much unless you understand the question and can translate this into the relevant/applicable data. Second, being capable of identifying patterns is a great ability to have.

What's the most common roadblock you hit in your work? How do you deal with it?

A key criterion for success when building self-driving solutions is to be able to mine data and correlate the transactional data with the user's decisions. In order to do so, it is crucial to secure access to as much data as possible from customers. However, that can sometimes be very hard as you will need to obtain the customer's consent -- and in some situations, the actual end user's consent -- that you are able to mine data and the corresponding actions associated with that data. Without that data, you cannot successfully train your algorithms.

To gain access, we are enhancing all our applications to explicitly ask for consent from individual users. From a software architecture perspective, it adds an additional layer of complexity to your data storage and data mining because you need to construct your algorithm in such a way that you filter data based on whether or not the user generating the data gave consent for the data to be mined.

Where is data analytics/data science headed in the next few years?

This is hard to say for someone in my profession, but I believe that data science will be automated in the future -- at least to some degree. Today, data scientists spend time analyzing and trying to understand data in the context of a requirement, to give insight. At the same time, we are building ever-smarter, self-learning algorithms to understand the data without human intervention -- essentially creating unsupervised learning.

This makes me pretty convinced that at some point in the near future we will see data science algorithms emerging that can learn to interpret big data to look for business-defined insights without the need for human intervention.