IT Strategy: The Cutting Edge of Analytics in Business

Screen-Shot-2014-09-10-at-12.24.33-PM.png

August 13, 2014 -- Clint Bidlack, President and CTO of ActivePrime, visited Babson University’s MBA students in the IT Strategy class taught by Professor David Charles. They discussed issues and trends in analytics, cloud computing and data quality. By far the most interesting topic for students was analytics. There’s a real thirst for understanding how analytics is being used in industry, what the major pitfalls are, and how to prepare oneself for leading successful analytics projects and analytics driven organizations. You might say that the “Moneyball Effect” is in full swing in MBA programs, where decisions are made based on data, not just instincts. Everyone wants better decisions and better outcomes by using big, real-time data.

We started with a few slides to introduce ActivePrime and some possible topics and then handed the reins over to the students. The resulting question and discussion session kept us on our toes.

One student asked: “Given the volume of big data being collected by my division, how do I even start to analyze the data?” We then discussed two paths, both driven from the first step of understanding what is trying to be answered, or being clear about what you want to know. Starting from the question, or set of expected questions to be answered is always a good practice. There’s nothing like setting a vision for what you want to accomplish to provide a focused direction to your analysis.

Path one is to build specific end user applications that answer these questions. Such applications tend to have value to a large number of users, and over an extended period of time, so that the investment in building the application is acceptable from an ROI perspective.

Path two is to build out more of more of a platform solution with providers that specialize in general purpose analytics tools and visualization. Providers like GoodData, Tableau or any other modern analytics expert, tend to be more expensive to manage over time, but they also provide more flexibility over the long term, allowing for a wide range of future, unforeseen questions to be considered.

Another great question revolved around preparing data for advanced predictive analytics, such as modern neural networks. This is a rather big topic that we only lightly touched on in the class. It has been estimated that more than 70% of the effort in machine learning deployments is due to data cleansing and data preparation work. Clearly data quality and data cleansing is important. ActivePrime wouldn’t exist otherwise! So getting your data right up front is important. The books we list at the end of this article describe in some detail how to plan for such data cleansing work.

After the class we presented the students with external resources for learning more about analytics in business. Our list included some great books to review and a video from a Google Tech Talk.

The video is a presentation of how advances in understanding how the brain perceives and processes sound and images can be translated into machine learning. The video covers the concept of deep learning algorithms, which teach machines how to interpret an “unlabeled” image from its base of “labelled” images. The strides made in this area are one of the drivers behind some big acquisitions by Google and Facebook in the past year.

The video is somewhat technical, but worth viewing because of the awe-inspiring nature of what these scientists are accomplishing. They have developed an algorithm that beats out almost all the expert, hand-built algorithms in the scientific community. For our MBA students, and anyone else involved in analytics, the reason they should watch this is because deep learning algorithms hold the promise of automating even more aspects of the analytics process, allowing for more rapid higher level analysis. This could also assist with partial automation of the data preparation stage. This would be a huge leap forward in analytics automation. And no doubt there are other applications of these algorithms that we just don't even see yet.

In addition, we recommended several books to the students that will give them insights and understanding into our topics. We hope you get a chance to expand your knowledge as well by reading them.

Book: Data Science for Business: Provost and Fawcett.

This is an excellent book to understand modern predictive analytics from a business perspective, but also gives enough technical information to allow one to converse with the technologists.

Book: Competing on Analytics; Davenport and Harris.

This is a high level business book. Chapter 2 is a well-known framework for assessing the current level of "analytics awareness/capabilities" within a company or group.

Book: Data Mining Techniques by Linoff and Berry

This book, revised in 2011, shows how data mining can be used to solve business problems. It can be a bit technical at times.

Don’t forget to subscribe to our quarterly update and get valuable information on CRM Best Practices, User Adoption, CRM On Demand Tips & Techniques, and our upcoming R26 series.